Google This is a digital copy of a book that was preserved for generations on Hbrary shelves before it was carefully scanned by Google as part of a project to make the world's books discoverable online. It has survived long enough for the copyright to expire and the book to enter the public domain. A public domain book is one that was never subject to copyright or whose legal copyright term has expired. Whether a book is in the public domain may vary country to country. Public domain books are our gateways to the past, representing a wealth of history, culture and knowledge that's often difficult to discover. Marks, notations and other maiginalia present in the original volume will appear in this file - a reminder of this book's long journey from the publisher to a library and finally to you. Usage guidelines Google is proud to partner with libraries to digitize public domain materials and make them widely accessible. Public domain books belong to the public and we are merely their custodians. Nevertheless, this work is expensive, so in order to keep providing this resource, we liave taken steps to prevent abuse by commercial parties, including placing technical restrictions on automated querying. We also ask that you: + Make non-commercial use of the files We designed Google Book Search for use by individuals, and we request that you use these files for personal, non-commercial purposes. + Refrain fivm automated querying Do not send automated queries of any sort to Google's system: If you are conducting research on machine translation, optical character recognition or other areas where access to a large amount of text is helpful, please contact us. We encourage the use of public domain materials for these purposes and may be able to help. + Maintain attributionTht GoogXt "watermark" you see on each file is essential for informing people about this project and helping them find additional materials through Google Book Search. Please do not remove it. + Keep it legal Whatever your use, remember that you are responsible for ensuring that what you are doing is legal. Do not assume that just because we believe a book is in the public domain for users in the United States, that the work is also in the public domain for users in other countries. Whether a book is still in copyright varies from country to country, and we can't offer guidance on whether any specific use of any specific book is allowed. Please do not assume that a book's appearance in Google Book Search means it can be used in any manner anywhere in the world. Copyright infringement liabili^ can be quite severe. About Google Book Search Google's mission is to organize the world's information and to make it universally accessible and useful. Google Book Search helps readers discover the world's books while helping authors and publishers reach new audiences. You can search through the full text of this book on the web at |http : //books . google . com/| Sci i^'PO. 36'0( 29-39) uiiiiuummuiiiiiiiiiiiiiiiiiiiiiBiiim HARVARD COLLEGE SCIENCE CENTER LIBRARY '^,^ jScv' S^ ^ »3 5^0 ^- (&wcv»^ SttBtttutUtti of ttsBliinatittt^ cjontributions from the mount wilson solar Observatory^ NO. 29 ANOMALOUS REFRACTION PHENOMENA INVESTIGATED WITH THE SPECTROHELIOGRAPH r F w. H. juuys Reprinted fronnthe Astropjtysicai Journal, Vol. XXVIII, December 190U '^ > ' \ •■ \ I t » /^^.y&. *G,..(?^<. ^?n«^ t "■■ Contributions from the Mount Wilson Solar Observatory, No. 39 Reprinted from the Aslropkysieal Jornnd^ Vol. XXVIII, pp 360-370, 1908 ANOMALOUS REFRACTION PHENOMENA INVESTI- GATED WITH THE SPECTROHELIOGRAPH By W. H. JULIUS According to the current interpretation of spectroheliograph results, dark flocculi indicate regions on the sun where the special gas, a line of which is used, exists in such conditions of density and temperature, that it strongly absorbs the light coming from deeper layers; whereas brighi flocculi show us regions where, in consequence of higher temperature or chemical or electrical causes, the radiation of the gas exceeds its absorbing effect. In a paper,' read before the Royal Academy of Amsterdam, in September 1904, I proposed an entirely different explanation of the same phenomena. A first attempt has there been made to account for the peculiar distribution of the light in photographs, secured with the spectroheliograph, by simply considering the anoma- lous refraction which waves from the vicinity of the absorption lines must suffer when passing through an absorbing medium, the* density of which is not perfectly uniform. If it proves possible to explain the observed facts on this basis, we shall be able to dispense with the assumption of any very marked differences as to the absorbing and emitting conditions of a certain gas or vapor in contiguous regions on the sun. Moreover, we then might assume the constituents of the solar atmosphere to be thoroughly mixed, their proportions in the mixture only varying with the distance from the sun's center. That our interpretation does not presuppose the separate existence of cloudlike masses of calcium or iron vapor or of hydrogen, looks like a simplification and, therefore, an advantage; but even if one were compelled, by other considerations, still to believe in the real existence of such separate luminous or dark accumulations of certain substances, it would nevertheless be necessary to consider the effect s Astrophysical Journal^ 3i, 278, 1905. 351] I 2 W, H. JULIUS which anomalous dispersion in those masses must have on the appearances revealed by the spectroheliograph. Among the advantages I derived from a visit to the Mount Wilson Solar Observatory in August 1907 was the opportunity of using the 5-foot spectroheliograph for some experiments on anomalous refrac- tion. It was expected that when light, coming from a source with a con- tinuous spectrum, traverses a space in which sodium vapor is unequally distributed, particulars about the distribution would be revealed by the spectroheliograph through the rejracHng properties of the vapor, rather than through its absorbing and emitting power. This expectation could be put to the test. As an equipment for the study of anomalous dispersion phenomena in sodium vapor, exactly similar to the one described in my paper on "Arbitrary Distribution of Light in Dispersion Bands,"* had already been secured for the Solar Observatory by Director Hale, the experiments were readily made, thanks to the laboratory facilities available on the mountain. The apparatus consists of a wide nickel tube, 60 cm long, the middle part of which is placed in an electric furnace, while the pro- jecting ends are cooled by jackets with flowing water. The tube contains a few grams of sodium, and is permanently connected to a Geryk pump to remove the air and the gases which escape from the sodium during the first stages of the heating process. An arrange- ment is provided by which density-gradients of various known directions and arbitrary magnitude may be produced in the sodium vapor. Sunlight coming from the mirror M (Fig. i) of the Snow telescope, which has a focal length of 60 feet, passes through the tube T on its way to the slit 5 of the spectroheliograph. The distance between T and S is about 560 cm. A lens Li gives an image of the sun near the middle of the tube T. P is a diaphragm, with an adjustable slit, of which the lens L^ projects an image in the plane of the diaphragm Q. Just behind the latter is a lens L^; in combination with L, this forms an image of a section of the tube in the plane of the slit S of the spectroheliograph. In this image (Fig. 2) the rectangular window I Astrophysical Journal, 25, 95, 1907. 352 ANOMALOUS REFRACTION PHENOMENA ^ \- a Fig. I V I 1 *^.J -.J Fig. 2 Hoc (H) ■A r ) Fig. 3 353 4 W, H, JULIUS of the caps of the tube' will of course come out with somewhat blurred edges, as only the middle section would show sharp. In A and B are projected the narrow nickel tubes for producing the required density-gradients. Their temperature may be varied at pleasure for this purpose by forcing an electric current or an air current through either of them. Cooling one of these tubes by an air current causes sodium vapor to condense on it; so in course of time drops of molten metal will hang on the tube, and fall off again. When a photograph is made, the first slit S of the spectroheliograph moves across the image in the direction of the arrow, and at the same time the second or camera slit moves across the photographic plate. Let us suppose the openings in P and Q (Fig. i) to be so adjusted, that the image of the slit in P exactly coincides with the slit in Q. Then all of the light which passes through P and traverses the vapor along straight lines is transmitted by Q, and therefore contributes to the intensity of the image of the tube section. Waves, however, that deviate so much in the sodium vapor as to be intercepted by the screen Q, will be absent from the spectrum of the transmitted light. K the furnace is slowly heated to 380° or 390°, the density of the vapor is pretty imiform in the middle part of the wide tube, and falls off toward the ends; but as the direction of the density-gradient" nearly coincides with that of the beam of sunlight, even the waves subject to anomalous dispersion will hardly deviate from the straight path. The D lines in the spectroheliograph retain nearly their normal appearance. If now we blow air through the tube 5, density-gradients are produced all around it in directions perpendicular to the axis of that tube. The D lines no longer show the same appearance through- out the field. In the spectrum of those parts of the field where per- ceptible gradients occur, the D lines then appear winged ; they are indeed enveloped in dispersion bands. As the width of these bands depends on the magnitude of the gradient, it will, in our case, vary along the lines, and reach a maximum at the place in the spectrum which corresponds to the plane passing through the axes of the tubes A and B, With increasing distance between 5 and B (Fig. 2) the width of the bands will diminish. « Compare Astrophysical Journal, 25, 97, 1907. Figs, i, a. 354 PLATE XLVI DeNSITY-GRAIllRNTS IN SoDlUM VaPOR AS SHOWN BV THE SrECTROHKLIOGKAPII ANOMALOUS REFRACTION PHENOMENA 5 Let us consider the monochromatic images of the tube-section produced by the spectroheliograph if the camera slit is set at different distances from the D lines. With the second slit at X 5850, outside the region of the dispersion band of D^, the illumination of the field is imiform (Plate XLVI, a) ; nothing is visible of the density-gradients existing around the cooled tube By because light of this wave-length travels along straight lines through the vapor. Proceeding to X 5870, we still are at such a distance from D, that the value of "X"=-R (^ representing the index of refraction, A the density of the vapor) is moderate. Steep gradients of the density are required to make the rays deviate sufficiently to miss the slit in Qy and such gradients are only to be found very near the surface of the tube B, We therefore obtain the image Py in which B appears surrounded by a narrow dark region. The third photograph, 7, was made with X 5877. For these waves the expression —^ is greater than for X 5870, so that smaller values of the gradient suffice to give the rays a perceptible incurvation. The result is a broader dark region all around £.' The photographs 8 and e were secured with the second slit on X 5881 and X 5885 respectively. This time the tube A was cooled instead of B, We see the dark aureole grow as the wave-length we are using approaches ^i>.=s890. Getting nearer still, the whole field would finally become dark. Similar results are obtained if we approach Di from the side of the greater wave-lengths, thus using waves for which —^ has increasing values. By a slight change in the arrangement of our experiment we may obtain the opposite effect, to wit, that merely rays suffering anomalous refraction enter the spectroheliograph, while the normally refracted light is prevented from reaching the slit. We have only to make the slit in P very wide, and to put a vertical bar (a match, for instance) in the middle of it, the image of which now falls exactly X In this image the lower right comer was cut off by a rubber tube accidentally crossing the path of the beam. 355 6 W. H. JULIUS on the slit in Q. Under these circumstances light, issuing from the divided opening in P, can be transmitted by Q only if it has been de- flected in the vapor. In this way the photographs f, ^, 6 were obtained, the second slit being set on X 5884, X 5886, X 5888 respec- tively. If there had been no density-gradients, the whole field would have shown dark; the bright regions, however, now prove the existence of the gradients. When taking K and 17, the tube j5, and when taking ^, the tube -4, was cooled. The following general statement is borne out by these experi- ments. If an illuminated absorbing vapor is investigated by means of the spectroheliograph, and the camera slit of the instrument is set on the edge of a dispersion band, marked irregularities in the brightness of the field will only appear at those places in the image which cor- respond to regions with large density-gradients in the vapor. Setting the slit nearer the middle of the dispersion band, we shall get evidence, in the image, also of the regions with smaller gradients, etc. Par- ticulars regarding the distribution of a vapor are thus dearly shown by the spectroheliograph through anomalous refraction, even in cases where the absorbing or emitting power of that medium would have failed to reveal its structure. The bearing of these inferences on astrophysical phenomena may now be considered a little more closely. Suppose we have a large mass of absorbing vapor of such average density that, if it were imiform, its absorption lines would appear rather narrow; and of such temperature, and condition of lumines- cence, that its emission lines are very faint. As soon as the density of this mass becomes irregular, some parts of it may give rise, when traversed by light from another source, to the appearance of dark or bright dispersion bands, greatly exceeding in width and strength its absorption or emission lines. It is therefore possible that anomalous refraction plays a very important part in the production of those phenomena which the student of astrophysics observes with his spectroscope or spectro- heliograph; we must inquire how far this is also probable. One might be inclined to object, for instance, that in our experi- 356 ANOMALOUS REFRACTION PHENOMENA 7 ments the ttse of a narrow and sharply limited source of light, placed at a fair distance behind the vapor, seemed to be a necessary condi- tion for obtaining any marked dispersion effects, and that in the sun similar circumstances are very unlikely to prevail. Indeed, the body of the sun, whatever the nature of the photosphere may be, is a large incandescent mass, closely surrounded by the absorbing vapors, so that the " source of light," if considered from a point of the chromosphere, subtends a solid angle of nearly 2ir. The reversing layer and the chromosphere have sometimes been compared to a thin, transparent layer of selectively absorbing varnish, covering a luminous (e.g., phosphorescent) globe: the photosphere. It seems very improbable that refraction in density-gradients of such a trans- parent envelope should be able to disturb to any perceptible degree the imiform brightness of that globe. The comparison, however, is entirely misleading, because, so far, an essential relation between absolute size and density-gradients is overlooked in it. But if carried through properly, it will lead us to the opposite conclusion, namely, that refraction in the solar atmos- phere must greatly alter the distribution of the light on the disk. If we wish to form an image, on a reduced scale, of the sun con- sidered as a refracting body, we have to reduce the radii of curvature of the rays in the same proportion as we do the diameter, for instance 10'^ times (so as to make the diameter of the photosphere 14 cm). By the general equation d^ I ds Rp ^ ^ we know that, for a given value of the refraction constant R, the radius of curvature /o of a beam of light is inversely proportional to the density-gradient j- in the direction toward the center of cmrature. In our image, therefore, the density-gradients have to be taken 10'® times as great as they are in the sun. Let us suppose that at a certain level in the solar atmosphere irregular density-gradients occur, that are of the same order of mag- nitude as the radial (vertical'i density-gradient in our earth's atmos- > Asirophysical Journal, 2$, Z07, 1907. 357 8 W. H. JUUUS phore, viz., i6Xio"".^ At the corresponding points in our image we then have to put 37 = 1 6. If the layer of "varnish" were really traversed by many density-gradients of this order of magnitude, it would be very different from ordinary transparent varnish, and cor- tainly be able to disturb the imiform brightness of the backgroimd, somewhat like a layer of glass beads or swollen sago grains. Even normally refracted waves would perceptibly deviate in an envelope of this kind. For if in our equation (i) we put -^=i?=o.s and ^ = 16, we get p=o. 125 cm, so that the average curvature of such rays is already sufficient for producing sensible changes in the diver- gence of beams on their way through a shell not thicker than o. i cm. Waves suffering anomalous refraction will of course be much more scattered by the same medium. Let us consider an absorbing substance which, at a certain level, occupies say only i per cent, of the solar atmosphere, taken as a perfect mixture. Its density-gradients will then be only ^-J-j- of those of the mixture. The refraction constant, on the other hand, for waves near one of its absorption lines may attain values as high as 1000 or 2000. With i{«=i6oo (observed in sodium vapor, Astrophysical Journal, 25, 108, 1907) our equation (i) becomes ^«^£fo''i6oop' In a level where in our image the irregular density-gradients of d^ the envelope were supposed to have an average value t"'*!^, the equation gives p=o.oo4cm. It is evident that imder such circumstances rays may easily deviate 90 degrees and more in the thin shell of transparent matter covering our globe, and thus give rise to a very imequal distribution of the light in photographs of it, secured with the spectroheliograph. > The frequent occurrence of density-gradients nearly perpendicular to the radii of the sun is rendered more probable still since increasing evidence has been obtained by Professor Hale of the existence of solar vortices, in which the convection currents (especially in sun-spots) are sufficiently strong to produce magnetic splitting of absorp- tion lines. (Cf. Nature, 78, 368-370, Aug. 1908.) 358 ANOMALOUS REFRACTION PHENOMENA 9 This conclusion holds quite as well with regard to the real sun. It follows directly from our only assumption that in some level of the sun there exist irregular density-gradients comparable in magni- tude with the vertical gradient in the earth's atmosphere. At lower levels greater gradients, at higher levels smaller gradients, may then be expected to prevail. As the validity of this assumption can hardly be doubted, we may infer that the existence of some important influ- ence of anomalous dispersion on astrophysical phenomena is not merely possible, but exceedingly probable^ in spite of the absence of narrow slits as sources of light. Although we are free to admit that the phenomena observed with the spectroheliograph on the solar disk are in part due to absorption and selective radiation, dependent on various conditions of tempera- ture or limiinescence, we may nevertheless inquire into some con- sequences to which one is led if only the effects of refraction in a mixture of vapors are considered. The composition of the solar atmosphere cannot be the same at all levels. As we get lower, the percentage of heavier molecules is likely to increase; but we should not presume too much as to the order in which the elements will come into evidence, on account of possible condensation, and because the pressure of radiation counter- acts gravitation to a degree that depends on the size of the particles, and, therefore, on numerous imknown conditions prevailing in the sun. Yet for each element a certain level must exist, in which its per- centage in the mixture is a maximum. Accordingly, the refracting properties of successive layers will be governed by different elements. A photograph, made with the spectroheliograph in a hydrogen line, shows a structure which of course depends on the distribution of all the hydrogen present in the successive layers, but is chiefly detei- mined by the density-gradients in a rather high level; whereas a photograph made with an equally strong iron line reveals especially the structure of lower regions. This explains the difference in char- acter between iron and hydrogen plates. It must be possible, on the other hand, to obtain almost identical photographs with different lines, provided they belong to the same 359 lO W. H. JULIUS element, or to elements that are most in evidence at about the same level of the sun; but then another condition has also to be fulfilled, viz., that the camera slit transmit rays of the same refrangibility in both cases. If, for instance. Fig. 3 represents the dispersion curve near Ha and near HB, the width and the position of the camera slit ought to be so chosen as to let in only waves corresponding to parts of the curve inclosed between equal ordinates in the two dispersion bands." Recently it has been found by Hale and Ellerman that, while Hfi and Hy and H^ lines give very similar results, photographs with the much stronger Ha line are widely different in some respects. Bright flocculi appear on these plates at points where no corresponding objects are shown by JETS. Moreover, the dark Ha flocculi, while showing a general agreement in position and form with those of JETS, are stronger and more extensive. In some instances, however, small areas appear dark in HS which are absent or fainter in Ha.* Such differences seem to be of the same character as those observed between photographs made with the slit in the broad calcium bands H or K at various distances from the central line. They may find a similar explanation if we assume that the average rays used in the Ha photographs were refracted to a higher degree than those used in the ifS photographs, but both by the same density-gradients. It is not improbable, therefore, that in the wings of Ha waves may be selected so as to give spectroheliograph results closely resembling HB plates. That also lines of different elements may give very similar results with the spectroheliograph, is exemplified by the case of calcium and iron. Among the beautiful collection of photographs secured on Mount Wilson I saw several iron (A 4045) plates resembling certain > Waves lying about symmetrically on either side of an absorption line, and answering the relation n— i — i — n' between the indices of refraction fi and n' of the medium for them, must give nearly the same spectroheliograph results on the greater part of the disk. This follows from a discussion of the various possibilities regarding the relative position of density-gradients and the source of light. Consequently an Hd plate, obtained with the camera slit centrally, so as to embrace the whole width of that rather narrow dispersion band, will scarcely differ, at first sight, from a photograph made with only one of the wings. • Memorie Soc. Spettroscopisti Italiani^ 37, 99, 1908. 360 ANOMALOUS REFRACTION PHENOMENA ii calcium (Hx) plates of the same daily series rather closely. As the alomk weight? of calcnnn and iron are not so very different and their levels of maximum density therefore probably not far apart, the refraction caused by these elements may bring out the density- gradients of nearly the same layer of the solar atmosphere. It will do so by a similar distribution of the h'ght in the two photographs — provided that rays of the same refrangibility are used in both cases. And this condition may be fulfilled by setting the camera slit on corresponding regions of the spectrum, in the manner illustrated by Fig. 3, if we imagine it now to bear on the calcium (H) line and the iron line. With a calcium and a hydrogen line such similarity could not be found. Far more evidence will of course be required before we shall be able to decide whether or not anomalous dispersion is the principal agent in determining the flocculent appearance of the solar disk. Plates secured with many lines of various elements should be com- pared. The powerful 30-foot spectrohdiograph of the "tower telescope" of Mount Wilson is excellently adapted for work of this kind, not only on account of its great dispersion, permitting the use of finer lines, but chiefly because it is provided with three camera slits, so that perfectly simultaneous photographs with different lines may be secured. By this arrangement, really comparable mono- chromatic pictures of the sun may be obtained, since the otherwise confusing influence of the variable refraction in our atmosphere is thus rendered harmless. I feel greatly obliged to Professor George E. Hale for having procured me the opportunity of making an investigation at the Moimt Wilson Solar Observatory, but more still for his keen and stimulating interest in the problems suggested by the application of the principle of anomalous refraction to astrophysics. I am also very much indebted to the kindness of Mr. F. Ellerman, Mr. W. S. Adams, and Dr. C. M. Olmsted for valuable information and assistance in connection with the inquiry here reported upon. Utrecht August 1908 361 T * : 1 •.^ 'VCv 6 J(/0.3 (Swn»^ itiBtttniiim of Wmlj^^m Contributions PROM the Mount Wilson Solar OBSERVATORY NO. 80 • ON T^E PROBABLE EXISTENCE OP A MAGNETIC FIELD IN SUN-SPOTS GEORGE E. HALE \ < J Reprinted from the AUmphytical JoHtnal^ Vol. XXVIII, November 1908 !•-'. '-H-A/J-. ihc^- •J^ ( , Contributions from the Mount Wilson Solar Observatory, No. 30 Reprinted from the Aslrophysical Jawnol, Vol. XXVin, pp. 315-343. IQ08 ON THE PROBABLE EXISTENCE OF A MAGNETIC FIELD IN SUN-SPOTS^ By GEORGE E. HALE The discovery of vortices surrounding sun-spots, which resulted from the use of the hydrogen line Ha for solar photography with spectroheliograph,* disclosed possibilities of research not previously foreseen. Photographs taken daily on Mount Wilson with this line suggest that all sun-spk>ts are vortices, and provide material for a discussion of spot theories which will soon be imdertaken. Reveal- ing, as they do, the existence of definite currents and whirls in the solar atmosphere, they afford the requisite means of testing the opera- tion in the sim of certain physical laws previously applied only to terrestrial phenomena. The present paper describes an attempt to enter one of the new fields of research opened by this recent work with the spectroheliograph. ELECTRIC CONVECTION In 1876 Rowland discovered that an electrically charged ebonite disk, when set into rapid rotation, produced a magnetic field, capable of deflecting a magnetic needle suspended just above the disk.^ It thus appeared, in accordance with Maxwell's anticipation, that a rapidly moving charged body gives rise to just such effects as are caused by an electric current flowing through a wire. Rowland's whirling disk therefore corresponds to a short wire helix, within which a magnetic field is produced when a current is passed through it. > A preliminary note bearing the title, "Solar Vortices and the Zeeman Effect, was sent to Nature for publication June 30. A brief abstract of this note appeared in Nature for August 20, together with a very interesting paper by Professor 2^many who was kind enough to examine some copies of my photographsi taken with the rhomb and Nicol in June. My own note was subsequently printed in Publications of ike Astronomical Society of ike Pacific^ ao, 220, 1908. « Hale, "Solar Vortices," Contributions from the Mount Wilson Solar Observatory, No. 26; Astrophysical Journal, a8, 100, 1908. 3 Rowland, " On the Magnetic Effect of Electric Convection," American Journal of Science (3), X5, 30, 1878. 363I I 2 GEORGE E. HALE Recent studies of the discharge of electricity in gases prove that gases and vapors, when ionized by one of several means, contain electrically charged particles. Moreover, at high temperatures carbon and many other elements which occur in the sun emit negatively charged corpuscles in great numbers; the complemen- tary positively charged particles must also be present, more or less completely separated from the negative corpuscles.^ Thus electro- magnetic disturbances on a vast scale may result from the rapid motions of charged particles produced by eruptions or other solar disturbances. Soon after the discovery of the vortices associated with sun-spots, it occmred to me that if a preponderance of positive or negative ions or corpuscles could be supposed to exist in the rapidly revolving gases, a magnetic field, analogous to that observed by Rowland in the labora- tory, should be the result. An equal number of positive and negative ions, when whirled in a vortex, would produce no resultant field,' since the effect of the positive charges would exactly offset that of the nega- tive charges. But Thomson's statement regarding the possible copious emission of corpuscles by the photosphere, and the tendency of negative ions to separate themselves, by their greater velocity, from positive ions, led to the belief that the conditions necessary for the production of a magnetic field might be realized in the solar vortices. Thanks to Zeeman's discovery of the effect of magnetism on radiation, it appeared that the detection of such a magnetic field should offer no great diflSculty, provided it were sufficiently intense. When a luminous vapor is placed between the poles of a powerful magnet the lines of its spectrum, if observed along the lines of force, appear in most cases as doublets, having components circularly polarized in opposite directions. The distance between the components of a given doublet is directly proportional to the strength of the field. As differ- ent lines in the spectrum of the same element are affected in different degree, it follows that in a field of moderate strength many of the lines may be simply widened, while pthers, which are exceptionally sensitive, may be separated into doublets. > J. J. Thomson, Conduction of Electricity through Gases, p. 165. > Unless separated by centrifugal force, as suggested by Professor Nichols. 364 MAGNETIC FIELD IN SUN-SPOTS 3 THE SUN-SPOT SPECTRUM It has long been known that the spectrum of a sun-spot differs from the ordinary solar spectrum in several particulars. If, for example, we examine the iron lines in a spot, we find that some of them are more intense than in the solar spectrum, while others are weaker. Again, we perceive that many of the spot lines are widened, and that the degree of widening varies for different lines. Finally, if the ob- servations are made with an instrument of high dispersion, it will be seen that some of the iron lines, which are single in the solar spectrum, are double in the spot spectrum. Such double lines were first seen by Young in 1892 with a large spectroscope attached to the 23-inch Princeton refractor. Walter M. Mitchell, who subsequently ob- served them with the same instrument, described the doublets as "reversals," which they closely resemble. Mitchell's papers contain a valuable series of observations of these "reversals" and other sim- spot phenomena.' Our previous investigations in this field on Moimt Wilson may be summarized as follows: I. The application of photography to the study of sun-spot spectra. A Littrow or auto-collimating spectrograph of 18 feet (5.5 m) focal length, used with the Snow telescope, gave good results, and permitted a great number of spot lines and bands, not previously known, to be recorded.* On the completion of the tower telescope last autumn, these observations were continued with a vertical spectrograph of 30 feet (9.1 m) focal length.^ Although the only grating available for work in the higher orders is a 4-inch (10 cm) Rowland, having 14,438 lines to the inch (567 to the mm), employed in my experiments in photographing sun-spot spectra at the Kenwood and Yerkes I Walter M. Mitchell, "Reversals in the Spectra of Sun-Spots," Aslrophysical Joumalf 19, 357, 1904; "Researches in the Sun-Spot Spectrum, Region F to a," ibid.y a a, 4, 1905; "Results of Solar Observations at Princeton, 1905-1906," ibid.f 34, 78, 1906. > Hale and Adams, "Photographic Observations of the Spectra of Sun-Spots," ConHbutions from the Mount Wilson Solar Observatory ^ No. 5; AsUrophysical Journal^ 23, II, 1906. 3 Hale, "The Tower Telescope of the Mount Wilson Solar Observatory," Con- iributions from the Mount Wilson Solar Observatory ^ No. 23; Astrophysical Journal, 27, 304, 1908. 365 4 GEORGE E. HALE Observatories,' the results secured with this instrument are very satis- factory, greatly surpassing those obtained with the i8-foot spectro- graph. They give the first photographic records of the "reversals" or doublets seen visually by Young and Mitchell, and reveal thousands of faint lines beyond the reach of visual observation. 2. The preparation of a photographic map of the sim-spot spectrum and a catalogue of all the lines. A preliminary map, consisting of 26 sections of 100 Angstroms each, covering the region A 4600-7200, was prepared last year by Mr. Ellerman from negatives made with the 18-foot spectrograph, and supplied to visual observers taking part in the sim-spot work of the International Solar Union. A much better map, to be made from negatives obtained with the tower telescope and 30-foot spectrograph, will be ready, it is hoped, within a year. The catalogue of lines, which involves a great amount of measurement for the determination of wave-lengths, is well advanced, and one section has been published by Mr. Adams.' 3. The identification of the numerous lines which constitute the fiutings in the spot spectrum. Photographs of the spectra of titanium oxide, magnesium hydride, and calcium hydride,^ made in oiu: labora- tory by Dr. Olmsted, have furnished the material for this purpose. The measurement of the lines in these fiutings is well advanced. 4. The interpretation of the change of the relative intensity of lines observed in passing from the solar spectrum to the spot spectrum. Investigations on the spectra of iron, manganese, chromium, titanium, vanadium, and other metals conspicuous in spots, made with the arc, spark, and flame, indicated that this change is due to a reduction of the temperature of the spot vapors.** Subsequent work with a new > Hale, "Solar Research at the Yerkes Observatory," Astrophysical Joumalj z6, 31 z, 1902. > Adams, " PreliminaTy Catalogue of Lines Affected in Sun-Spots, Region X 4000 to X4500," CorUrihutums from the Mount Wilson Solar Observatory, No. 22; Astro- physical Journal, 37, 45, 1908. 3 Olmsted, "Sun-Spot Bands Which Appear in the Spectrum of a Calcium Arc Burning in the Presence of Hydrogen," Contributions from the Mount Wilson Solar Observatoryy No. 21; Astrophysical Journal, 37, 66, 1908. 4 Hale, Adams, and Gale, " Preliminary Paper on the Cause of the Characteristic Phenomena of Sun-Spot Spectra," Contributions from the Mount Wilson Solar Obser- vatory, No. 11; Astrophysical Journal, 34, 185, 1906; Hale and Adams, "Second Paper on the Cause of the Characteristic Phenomena of Sun-Spot Spectra," Contribu- tions from the Mount Wilson Solar Observatory, No. 15; Astrophysical Journal, 35, 75. 1907- 366 MAGNETIC FIELD IN SUN-SPOTS $ electric furnace by Dr. Kling,^ the details of which have not yet been published, seems to leave little doubt that this explanation is correct. It is supported by the presence in the spot of compounds which appear to be dissociated at the higher temperature outside the spot, and by the resemblance of spot spectra to the spectra of red stars.* While our investigations have thus furnished a plausible explana- tion of some of the characteristic phenomena of sun-spot spectra, the widening of lines and the presence of doublets are among the remaining peculiarities that demand consideration. As we have seen, however, these very peculiarities are precisely what would be expected if a magnetic field were present. Prompted by the theoretical con- siderations outlined above, and encouraged by their apparent agree- ment with the facts of observation, I decided to test the components of the spot doublets for evidences of circular polarization and to seek for other indications of the Zeeman effect. METHOD OF OBSERVATION The tower telescope forms an image of the sun, about 6 . 7 inches (17 cm) in diameter, on the slit of a vertical spectrograph, of 30 feet focal length. This instrument, to which reference has already been made, stands in a well with concrete walls, the grating being about 26J feet (8 m) below the surface of the groimd. The temperature at the bottom of the well is so constant that exposures of any desired length may be given, without danger of a shift of the lines resulting from expansion or contraction of the grating. A Fresnel rhomb and Nicol prism^ are moimted above the slit, so that the light of the solar image passes through them. If the doublets in spots are produced by a magnetic field, the light of their components, circularly polarized in opposite directions, should be transformed by the rhomb into two > King, "An Electric Furnace for Spectroscopic Investigations, with Results for the Spectra of Titanium and Vanadium," CofUributions from the Mount Wilson Solar Observatory, No. 28; Astrophysical Journal, 28, 300, 1908. 9 Hale and Adams, "Sun-Spot Lines in the Spectra of Red Stars," Contributions from the Mount Wilson Solar Observatory, No. 8; Astrophysical Journal, 33, 400, 1906; Adams, "Sun-Spot Lines in the Spectrum of Arcturus" Contributions from the Mount Wilson Solar Observatory, No. 13; Astrophysical Journal, 34, 69, 1906. 3 Obtained for this purpose in 1905, when the idea of searching for the 2^man effect in sun-spots had already occurred to me. A visual test of the spot lines for plane polarization, made with the 18-foot spectrograph in 1906, before we had photo- graphed the doublets, gave negative results. 367 6 GEORGE E. HALE plane polarized ra3rs, differing 90^ in phase. Thus, in a certain posi- tion of the Nicol, the light from the red component should be trans- mitted and that of the violet component cut off. When rotated 90° in azimuth, the Nicol should transmit the violet component and cut off the red component. Complete extinction of either component is hardly to be expected, because the light from the spot does not, in general, come exactly along the lines of force, and the doublets may therefore exhibit some traces of elliptical polarization. Moreover, the beam of simlight imdergoes two reflections on the silvered surfaces of the coelostat and second mirrors of the tower telescope, where elliptical polarization must again be introduced.' By setting the rhomb at the proper angle, the latter effect, which is not very large, can be almost wholly eliminated, but the former may play some part, even when the spot is at the center of the sim. The light of the spot, after transmission through the rhomb and Nicol, comes to a focus in the plane of the slit. While photographing the spot spectrum the slit is covered, except at its central part, where a portion corresponding in length (from i to 2 mm) to the diameter of the umbra, receives the light. During the exposure, which may continue from a few minutes to over an hour, the image of the umbra is kept as nearly as possible central on the slit, any irregularities in the motion of the driving-clock being corrected by the observer. As the exposure for the spot spectrum is from five to twenty times as long as for the solar spectrum, it is evident that care must be taken to prevent light from regions outside the spot from entering the slit. For a comparison spectrum sunlight is used, generally from a point in the solar image a short distance away from the spot, where none of the characteristic spot phenomena appear. During the exposure, that part of the slit which previously received the light of the umbra is covered, and sunlight admitted on either side. The light of the comparison spectrum passes through the rhomb and Nicol, both of which occupy the same positions as in the case of the spot. Care is taken to see that the grating is fully illuminated, both for the spot and comparison spectra, in all positions of the Nicol. > A study of the elliptical polarization of these mirrors has been made by Dr. St. John. 368 1I .»! 11 ii MAGNETIC FIELD IN SUN-SPOTS 7 CntCULAR POLARIZATION ALONG THE LINES OF FORCE My first observations were made on June 24, in the second cMrder of the grating, but the results were not conclusive. On June 25 I obtained some good photographs, in the third order, of the region A 6000-6200, using Seed's "Process" plates, sensitized for the red by Wallace's three-dye formula." These clearly showed a reversal of the relative intensities of the components of spot doublets when the Nicol was turned through an angle of 90*^. Moreover, many of the widened lines were shifted in position by rotation of the Nicol, indi- cating that light from the edges of these lines is circularly polarized in opposite directions. The displacements of the widened lines ap- peared to be precisely similar in character to those detected by Zeeman in his first observations of radiation in a magnetic field. A series of photographs, made with the Nicol set at various angles, soon showed the two positions giving the maximum eflfect. At these positions the weaker components of the strongest doublets are not always completely cut oflF, but their intensities are greatly reduced. Sometimes hardly a trace of the weaker component remains, as may be seen in the case of the vanadium doublet at X 5940 . 87 (Plate XLVII). In this plate No. 5 shows the doublet in the ordinary spot spectrum, photographed without the rhomb and Nicol. No. 4, from a photograph (T 190) made with the Nicol set at 61® E., shows only the red component of the doublet. No. 3 illustrates the effect of turning the Nicol 90®: only the violet component remains. Other spot lines in these photographs change in a similar way. Photographs like these seemed to leave no doubt that the compo- nents of the spot doublets are circularly polarized in opposite directions. Since the only known means of transforming a single line into such a doublet is a strong magnetic field, it appeared probable that a sim- spot contains such a field, and that the widening and doubling of the lines in the spot spectrum result from this cause. But much remained to be done before the proof could be regarded as complete. In the first place, it was necessary to make sure that the displace- ment of the lines other than doublets ^as not due to instrumental causes, such as a change in the illumination of the grating produced by rotating the Nicol. As already stated, care was always taken to < Astrophyskal Journal^ 26, 299, 1907. 369 8 GEORGE E. HALE see that the ruled surface was filled with light before making an expo* sure. Moreover, the magnitude of the displacement was much greater for some lines than for others, and the fact that the shifts were determined with respect to lines of the solar spectrum, whose light had traversed almost the same path as that of the spot in rhomb and Nicol, seemed to leave little room for doubt as to their true character. How- ever, a rigorous test could be applied. The spot spectrum, as well as the solar comparison spectrum, is crossed by lines due to the absorp- tion of water vapor and other gases in the earth's atmosphere. If a change of illumination due to the rotation of the Nicol were concerned, these lines should be displaced from their normal positions. But no such shifts were observed. Furthermore, it is known that the lines of most flutings are not affected by a magnetic field. Accordingly, the cyanogen fluting at X 3883 was photographed in the spot spectrum, with the Nicol set in two positions 90*^ apart. Three lines in this fluting, which I have measured on negative T 132, made in the fourth order, show a mean displacement of 0.0004 Angstrdms, with respect to the corresponding lines of the solar comparison spectrum. This quantity is well within the error of measurement." We may therefore conclude that the Nicol displaces only those lines which show polari- zation phenomena. While measuring this plate, and others taken in the more refran- gible part of the spot spectrum, it was foimd that few of the lines in this region show large shifts. A group of doublets was encountered near X 4400, the components of which are circularly polarized in opposite directions. In general, however, the shifts produced by rotating the Nicol decrease from the red toward the violet end of the spectrum. Since this preliminary work I have made over two hundred photo- graphs of spot spectra with polarizing apparatus before the slit. In addition to this collection of plates, numerous photographs of spot spectra, some taken with polarizing apparatus by Dr. St. John, and others made without Nicol or rhomb by Mr. Adams and myself, are available for study. These have been used for the investigation described in the following pages. s The head and several lines of the titanium oxide fluting at X 5598, which have since been measured by Mr. Adams, also show no displacement when the Nicol is rotated. 370 PLATE XLVni :908, -September 9, 6" 20"' a. M. Scnle: Suns Uiametert^o.3 Meter PLATE XLIX I Flocc(pi.i, Showjnc Right- and Left-Handed Vortices igo8, September 7, 6'' lo"' a, m. Scale: Sun's Diameter=o.3 Meier MAGNETIC FIELD IN SUN-SPOTS 9 BEVEKSED POLARITIES OF RIGHT- AND LEFT-HANDED VORTICES A second test, which also bears upon the hypothesis that the field is produced by the revolution of electrically charged particles in the spot-vortex, may now be described. If a Nicol is set so as to cut oflf the violet component of a doublet observed along the lines of force of a magnetic field, reversal of the current will cause the red component to disappear and the violet component to become visible. Reversal of the direction of the current in a magnet corresponds to reversal of the direction of revolution in a solar vortex. If it could be shown, by an independent method, that in two sun-spot vortices the charged particles are revolving in opposite directions, the red components of the doublets should appear in the spectrum of one spot, and the violet components in that of the other, the position of the rhomb and Nicol remaining unchanged. Fortunately, the spectroheliograph plates indicate the direction of revolution in the solar vortices. The vortices are constantly changing in appearance, and the stream lines are not always clearly defined. Plates XLVni and XLIX are reproduced from photographs of the sun made by Mr. Ellerman with the s-foot spectroheliograph on September 9 and 10. They show two spots, one in the northern, the other in the southern hemisphere, with vortices indicating revolution in opposite directions, if we may judge from the curvature of the stream lines.' Portions of the spectra of these spots, photographed by myself on September 9, are reproduced in Plate XLVII. No. i shows the spectnun of the southern spot, in which the direction of revolution was clockwise, taken with the Nicol set at 29® W. Only the red components of the doublets appear. The northern spot, in which the revolution was counter-clockwise, was then photographed (2). Although the Nicol and rhomb remained in the same position as before, the red components of the doublets are now cut off, while the violet ones are visible. During this exposure the slit was kept on the western umbra of the northern spot, which was divided into two parts by a bridge (not shown in the reproductions) . Another exposure, with Nicol and rhomb as before, was then made on the eastern umbra of the same spot (3), with results similar to those obtained for the western umbra. For the final exposure (4) the slit was kept on the eastern > Right- and left-handed vortices have also been found in the same hemisphere . 371 lo GEORGE E, HALE umbra of the northern spot, and the Nicol rotated 90^. As was to be expected, the red components were brought into view, and the violet components extinguished. This spectrum is therefore precisely similar to that of the southern spot, which was taken with the Nicol in the reverse position. This result has been confirmed by other photographs, which indi- cate that the direction of the displacement always depends upon the direction of revolution in the vortex. If this relation is found by futiure observations to hold generally, we may conclude that the field is always produced by the revolution of particles carrying chaiges of like sign. PLANE polasizahon across the lines of force So far we have confined oiur attention to polarization phenomena observed along the lines of force. But it is well known that the doub- lets are, in general, transformed into triplets, when observed in a magnetic field at right angles to the lines of force. The components of the triplets are plane polarized, the central line in a plane at right angles to the plane of polarization of the side components. It should be possible to detect similar phenomena in spot spectra, if they are produced in a magnetic field. It naturally happens that these spectra are most commonly observed when the spots are not very far removed from the center of the sun, because foreshortening near the limb reduces the umbra to a narrow strip difficult to keep on the slit. This may partially explain why oiur photographs of spot spectra, taken without polarization apparatus, show the doublets without a trace of a central component. But it does not account for the failure of the central line to appear in the spectra of spots well removed from the center. It is true that a few triplets occur in all of our spot spectra, such as A 5781.97, A 6064.85, and A 6173.55. But these I have regarded as probable examples of an exceptional type of lines, observed in the laboratory as triplets along the lines of force. Mitchell records certain cases in which many spot doublets were seen as triplets,' but he also notes the existence of doublets in the spectra of spots near the limb.' In X Astrophysical Journal, 24, 80, 1906. • Ibid,, xg, 357, 1906. 372 PLATE L 5436-80 '5) 60- E. {3) Spror 1908, Sipt. ■ MAGNETIC FIELD IN SUN-SPOTS ii one interesting observation described and illustrated by Mitchell, the lines appeared double across the umbra and one side of the penumbra, while on the other side of the penumbra they changed into triplets.' Since the beginning of my work on the Zeeman effect in the sun, there have been few opportunities to observe the spectra of spots near the limb. These I have utilized, not in attempting to photograph the triplets (which will be tried later), but in testing the polarization phenomena of the spot lines. The rhomb was removed, and the Nicol employed alone. At right angles to the lines of force the Nicol, when in a certain position, should cut off the outer components of a triplet or the edges of a widened line. In another position, 90^ distant, the central component should be extinguished, and the outer com- ponents or edges transmitted. Thus, in the second case, lines which are not too diffuse should be photographed as doublets, while in the first case the central component should appear alone. Plate L reproduces some photographs of a spot near the west limb, made on September 14. The seeing was poor, and neither the Ha image nor the spectra are sharply defined. In Fig. i, Plate L, (i) and (s) represent the solar spectrum; (2) the spot spectrum, photographed with the Nicol set at 60® E.; (3) the spot spectrum, with Nicol set at 60® W. ; (4) the same region of another spot spectrum, photographed near center of sim without Nicol." At 60® E. the Nicol cuts out the central line of X 5436. 80, while at 60® W. it transmits this line and cuts off the side components. Other settings of the Nicol gave the following results, which appear on the same negative (T 200) : 90® E., single; 30*^ E., double; o®, single, but wide; 30° W., single, but wide. Other photographs, made in this and other regions of the spectrum, gave similar results, many lines being narrow in some positions of the Nicol and wide in others. Only one case (X 5436. 80) of imdoubted doubling of the lines has been found. The short time available for work, under favorable atmospheric conditions, when a sufi&ciently large spot was near the limb, prevented the observations from being carried farther. X Ibid,, 24, 80, Z906. * This cut is very unsatisfactory, but a better one could not be obtained. 373 12 GEORGE E. HALE LABORATORY TESTS If the widened lines and doublets in spot spectra are produced by a magnetic field, an equal degree of widening and an equal separation of the components of doublets should be found in the laboratory when the same lines are observed in a field of equal strength. As the neces- sary apparatus was fortunately available, the work was at once under- taken in our Pasadena laboratory by Dr. King. A brilliant spark is produced by a high potential transformer between electrodes sup- ported in the field of a large Du Bois magnet. The light, passing through the pierced pole-pieces, falls on a lens, which forms an image of the spark on the slit of a vertical spectrograph, after reflection on a mirror mounted at an angle of 45° above the slit. This spectrograph, which is precisely similar to the 30-foot spectrograph used with the tower telescope, also stands in a constant temperatiure well, with the slit about three feet above the floor of the laboratory.' It may be used as an instrument of 30 feet focal length, or, as in the present case, a s-inch (13 cm) objective of 13 feet (4 m) focal length, with a 5-inch plane grating, having 14,438 lines to the inch (567 to the mm), can be swung into the axis of coUimation 13 feet below the slit. With this shorter focal length the dispersion in the second or third order of the grating is amply sufficient for the present purpose. If all of the doublets observed in spot spectra could be photo- graphed in the laboratory, it would be easy to make a satisfactory comparison. Unfortunately, however, most of these lines are very faint in the spark, and as the great majority of them occur in the less refrangible part of the spectrum, exposures of from fifteen to twenty TABLE I Iron Doublets Wave-Length AA.Spaxk AX, Spark 5.1 AX, Spot a AX. Spark AA, Spot 6213.14 6301.72 6302.71 6337-05 0.703 0.737 1.230 0.89s 0.138 0.144 0.241 0.175 0.136 0.138 0.252 0.172 —0.002 —0.006 +0.011 -0.003 5-2 5-3 4.9 5a z Hale, "The Pasadena Laboratory of the Mount Wilson Solar Observatory," CofUnbuiions from the Mount Wilson Solar Observatory, No. 27; Astrophysical Jour- nal, 28, 244, Z908. 374 MAGNETIC FIELD IN SUN-SPOTS 13 hoiirs axe sometimes required to bring out even the stronger doublets. The results hitherto obtained for the iron doublets are brought together in Table I. I am indebted to Mr. Adams for these measures and for many of the others given in this paper. Miss Burwell and Miss Wickham have also assisted in the measurement of the spot and spark photographs. The first column gives the wave-length of the doublet; the second, the separation in Angstr5ms of the components, observed along the lines of force in a field of about 15,000 gausses;' the third, the quantity given in colimm 2 divided by 5.1; the fourth, the separation of the components observed in the spot spectrum; the fifth, the residuals obtained by substracting the quantities in the third colunm from those in the fourth; the last colunm gives the ratio of the separation in the spark, for a field of about 15,000 gausses, to the observed separation in the spot. The mean value of this ratio, 5.1, gives an approximate measure of the strength of the field in spots, which comes out about 2900 gausses. The agreement between the spot and laboratory results is so close that it can hardly be the result of hance. But when we come to the case of titanium, observed in the laboratory in a field of about 12,500 gausses, we find a very different condition of affairs. TABLE II TlTANIXTM DOXTBLKTS WaTe-Lenstk AA,Spuk Ak, Spark AA,Spot a AA, Spaik AA, Spot 5903.56 5938.04 6064.85 6303.98 63x2.46 0.732 0.737 0.876 0.493 0.615 0.144 0.145 0.172 0.097 O.Z2I 0.086 0.080 0.184 0.093 O.09X -0.058 -0.065 + O.OZ2 — 0.004 -0.030 8.5 9.2 4.8 5-3 6.8 If we use the factor 5 . i employed in the case of iron, we find that two of these doublets, X 6064.85 and A 6303. 98, agree closely in spot and spark. In some of our spot photographs X 6064.85 appears to be a triplet, though the components are not clearly separated. With the > This value of the field strength may be in error by zooo gausses, because of the disturbing effect of the iron electrodes, 375 14 GEORGE E. HALE rhomb and Nicol a faint central component persists when either the red or the violet component is cut off. It is possible that this central line is due to some substance other than titanium in the spot, but it is certainly very nearly in the position of the solar titanium line/ A 63 1 2 . 46 gives a residual of o. 03 AngstrSms, which exceeds the error of measurement. The other doublets, A 5903. 56 and A 5938.04, show in the spot spectrum but little more than one-half the separation that would be expected on the assumption that the strength of the field is the same for all of these lines. On consideration it will be seen, however, that the separation of the doublets must depend, in some degree, on the distribution of the absorbing vapor in the solar atmosphere, and on the coefficient of absorption of the particular line employed. A striking instance of this kind, affecting lines of the same series, is illustrated in the case of hydrogen, described in a previous paper.* Although the Hh line extends to the upper part of the chromosphere and prominences, the mean level represented by its absorption is much lower than that given by Ha. The consequence is that Ha enables us to photograph the solar vortices, the characteristic stream lines of which do not appear at the lower Hh level. Similarly, if the intensity of a given titanium line falls off rapidly, the level represented by this line may be com- paratively low. If, on the other hand, its intensity curve is of such a form as to indicate that the absorption at higher elevations plays an important part, the mean level represented by the line may be considerably higher than in the previous case. To settle this question we must know: (i) The range of elevation in the spot of the vapors of iron, titanium, and other elements; (2) the intensities of the lines of these elements at different levels; (3) the rate at which the strength of the field decreases upward. In the absence of information regarding the first two points, we may inquire as to the probable relative behavior of titanium, iron, and other elements if the distribution of the vapors at different levels were the same as in the chromosphere. From a discussion of a large > It is conceivable that under conditions analogous to those that give rise to the H3 and K3 lines, a doublet might be produced within the strong magnetic field of the spot, and a single line, at the center of the doublet, by the absorption of the vapor at a high level, where the field strength is low. « Solar VorliceSf p. 3. 376 MAGNETIC FIELD IN SUN-SPOTS 15 number of photographs of the flash spectrum, made by different observers at several eclipses, Jewell has compiled a table showing the heights above the sim's limb attained by various lines in the blue and violet.* The heights for titanium range from 100 miles (160 km) for A 4466.0 to 3500 miles (5640 km) for A 4466. 7, while certain strong enhanced lines in the ultra-violet reach elevations of 6000 or 8000 miles (9660 or 12,880 km). For iron the minimum height is 200 miles (320 km) for A 4482. 4 and the maximum 1000 miles (1610 km) for A 4584.0. Chromium ranges from 100 miles for A 4280. 2 to 1200 miles (1930 km) for A 4275.0; manganese from "100 miles or more" for A 4451.8 to "800 miles (1290 km) or more" for A 4030.9; vanadium from 100 miles for X 4390 . 1 to 200 miles for A 4379 . 4. It thus appears that the range in level represented by the titanium lines is much greater than for the lines of iron, chromium, manganese, and vanadium. If the vapors were similarly distributed in spots, the maximum strength of field indicated by the titanium lines should therefore correspond with the maximum value for iron, but some titanium lines, produced by absorption at higher mean levels, should give lower field strengths. Chromium should agree more nearly with iron. Vanadium, if the less refrangible lines reach no greater elevations, should give closely accordant (maximum) values for the field strength. It will perhaps be possible, with the aid of the 30-foot spectrograph, to determine the relative levels in the chromosphere attained by most of the lines in question, but it is a much more difficult matter to do this for sim-spots. I hope, however, that our new spectroheliograph of 30-feet focal length may throw some light on this subject. It is evident that these considerations will have no bearing on the present problem, unless the field strength decreases very rapidly upward in spots. That this probably occurs is shown by the fact that the D lines of sodium and the b lines of magnesium are usually but slightly affected in the spot spectrum,' and are displaced through a very small distance when the Nicol is rotated. Thus, at the level represented > "Total Solar Eclipses of May 28, 1900, and May 17, 1901/' Pyhlications of the U, 5. Naval Ohservaiory, Second Series, Vol. IV, Appendix I. • Except for the strengthening of the wings» which may be produced by some cause other than a magnetic field. 377 i6 GEORGE E. HALE by these lines, which attain elevations in the chromosphere probably not exceeding 5000 miles, the field strength is reduced to a small fraction of its maximum value. The following doublets have been measured in the spectrum of chromium: TABLE III Chroioxtm Doublets AA, Spaik AA, Spuk Wftme-Leagth AA.S|»ik 4.9 AA.Spot < AA.Spot 5304.36 0.636 0.130 0.188 + 0.058 3-4 5387.16 0.676 0.138 0.085 -0.043 8.0 5713 00 o.6zo O.Z94 o.z6x + 0.037 3-7 5781 . 40 0.755 0.154 O.Z9Z -0.033 6.9 5781-97 0.999 O.Z88 0.9Z9 + 0.094 4.3 5783 29 0.779 0.158 0.137 — 0.091 5-6 5784.08 0.790 0.147 O.I9Z — 0.096 6.0 5785- 19 0.707 0.144 0.137 — 0.007 5.1 In photographing these lines in the spark, the strength of the field was 12,500 gausses. The strength of the field in spots, as indicated by the mean separation of the chromium doublets, is therefore 2600 gausses. The above tables comprise aU of the doublets hitherto observed both in spots and in our laboratory. It was at first hoped that the shifts of lines, on photographs of the spot spectrum made with the rhomb and Nicol, would serve as satisfactory data for comparison with laboratory results. But when the small magnitudes of these shifts, and the wide differences in the character of the lines were taken into accoimt, it appeared that comparisons based on such data could have but little weight. When a line is clearly resolved into a doublet, rotation of the Nicol cuts off the right-handed or left-handed light, and produces a shift equal to the separation of the components. But when the strength of the field is only sufficient to widen a line, that portion of the widened line where the right-handed and left-handed components overlap is composed of ordinary unpolarized light, not affected by rhomb or Nicol. If the components are narrow, this region may also be nar- row. But if they are broad, only the outer edges of the components will be cut off when the Nicol is rotated. 378 / MAGNETIC FIELD IN SUN-SPOTS 17 If a magnetic field is the principal cause of the widening of lines in spots, their widths should be roughly proportional to the separation of the components of the corresponding doublets observed in a field of equal strength. Bearing in mind the differences in the character of the lines, and the probable effect of variations in the mean level of absorption, we can hardly expect a very dose agreement. But some evidences of relationship should appear, if a magnetic field is present. In the following tables the widths of various iron lines are compared with the separations of their components in the spark. To facilitate the comparison, the distances between the centers of the components, photographed in a field of about 15,000 gausses, are divided by 2 . 9, which reduces them to approximate equality with the widths in spots. TABLE IV Wn>THs oj Ikon Lines in Spots Wft^e-Lencth AA, Spark AA, Spark a. 9 Width in spots < 6x36. 19 0.38 0.13 0.15 + 0.02- 6137.92 0.50 0.17 o.x6 — O.OI 6x91 . 78 0.43 0.15 0.14 — O.OI 62x9.49 0-59 0.20 0.23 +0.03 6246.54 0.67 0.23 0.24 + O.OX 6252.77 0.45 o.x6 015 — O.OI 6265.35 0-55 0.X9 0.20 +0.01 631552 0.59 0.20 o.x6 —0.04 Enhanced line 6318.24 0.40 0.X4 O.X4 0.00 6335 -55 0-55 0.19 0.20 +0.0X 6393.82 0.46 0.16 0.18 +0.02 6400. 22 0.58 0.20 0.22 +0.02 641X.86 0.56 0.X9 0.17 —0.02 6417.13 0.69 0.24 0.15 — 0.09 Enhanced line 6420. X7 0.57 0,20 0. X9 — o.ox 642X . 57 0.64 0.22 o.x8 —0.04 643107 0.54 0.X9 0.19 0.00 6456.60 o-SS 0.19 0.22 + 0.03 Enhanced line 6495" 0.54 0.X9 o.x8 — O.OI The exceptionally large residuals of the enhanced lines may be due to the fact that the weakening of these lines in spots makes them very difficult to measure. But it is perhaps possible that another cause may account for the negative sign of most of their residuals in Tables IV and VI. Assume that in the lower part of spots the field is most intense and the reduction of temperature most marked. In 379 i8 GEORGE E. HALE TABLE V Widths of Ikon Lines in Spots AA, Spuk Wave-Length AX, Spark a.9 Width In Spots I 5083.58 0.41 0.14 O.X5 + O.OX 5098 .88 0.42 0.14 0x3 — o.ox 5107 .63 0.2s single* 0.09 o.xx +0.02 5107 .82 O.X4 5x10 57 0.45 0.15 0.X7 +0.02 5"3 90 single 0.09 5"5 30 0.41 0.Z4 0.09 —0.05 5x27 53 0.51 0.18 O.X5 -0.03 5137 56 0.45 0x5 0.13 —0.02 5139 43 0.56 0.19 o.x6 -0.03 5139 .64 0.51 0.18 0.15 —0.03 5143 .11 0.42 O.X4 o.xx -0.03 5162 45 0.47 o.x6 0.X4 —0.02 5167 .68 0.35 0.12 O.X2 0.00 5171 78 0.39 0x3 O.X5 +0.02 5191 63 0.57 0.20 o.x8 —0.02 5192 52 0.56 0.X9 0.17 —0.02 5195 .11 0.33 o.xx 0.X4 +0.03 5198 89 single O.XO 5208 78 0.48 o.z6 O.X4 —0.02 5215 35 0.45 0.15 0.15 0.00 5216 44 0.23 0.08 0.12 +0.04 5"7 55 0.47 0.16 0.15 —o.ox 5227 04 0.47 0.16 0.19 +0.03 5217 36 0.32 o.xx 0.15 +0.04 5230 03 0.50 0.17 0.15 —0.02 5233 12 0.38 0x3 0.14 + O.OX 5343 66 0.29 O.IO o.xx + O.OX 5250 82 0.49 0.X7 0.X4 -0.03 5203 49 0.47 o.x6 O.X3 -0.03 5266 74 0.38 0.13 O.X4 + O.OX 52^ 72 0.39 0.X3 0x5 +0.02 5»73 56 0.53 o.x8 o.xx —0.07 5276. 17 0.31 o.xx O.X2 + O.OX 5281. 97 0.44 0.15 O.XX —0.04 5283 .80 0.49 0.X7 O.X4 —0.03 5302 48 0.48 0.X7 O.X7 0.00 5316 79 0.32 o.xx O.XX 0. 00 Enhanced tine 5324. ■37 0.48 0.16 o.x6 0.00 5328 24 0.37 0.13 0.17 +0.04 5328. 70 0.49 0.X7 0x3 — 0.04 5340 22 0.48 0.16 o.x6 0.00 5365 07 0.30 O.XO O.XO 0.00 5367 67 0.31 o.xx 0.12 +0.0X 5370 17 0.36 0.12 O.X2 0.00 5371 73 0.33 o.xx o.x6 +0.05 5383 58 0.37 0x3 0.X3 0.00 5393 38 0.52 0.18 o.x8 0.00 5397 34 0.48 o.x6 0.20 + 0.04 5400.71 0.42 0.14 o.xx -0.03 • "Single" in thcM Ubiet does not it was not clearly separated on the plate first order. that the Hne is not affected by the field, but nciely that Several of these photographs were made tai the 380 MAGNETIC FIELD IN SUN-SPOTS 19 TABLE V^Cantinued WftTe-I^Migth ilA, Spark AA, Spaik 3.9 Width in Spots < 5404.36 0.38 0.13 0.16 +0.03 5405.99 0.23 0.08 0.15 +0.07 5411.1a 0.40 0.14 0.13 — O.OI 5415.42 0.38 0.13 0.15 +0.02 5424.29 0.40 0.14 0.15 +0.01 5439.92 0.48 0.16 0.16 0.00 5434.74 single O.IX 5447.13 0.51 0.18 0.19 + O.OX 5455.83 single 0.20 consequence of the reduced temperature, the enhanced lines are greatly weakened. Hence an unusually large proportion of the absorption which gives rise to these lines may occur at greater elevations, where the temperature is higher and the field weaker. In this case, the field intensities indicated by the enhanced lines should be below the average value. In view of the fact that the rate of change of intensity with level is not the same for all lines, it is evident that many more cases must be included in any satisfactory test of this h3^thesis. From the same course of reasoning it follows that lines which are most strengthened in spots should, in general, be most widened. This appears to be true, but a careful quantitative comparison will be made, both for strengthened and weakened lines, and published in a subse- quent paper. It must not be forgotten that a considerable increase of temperature in the higher spot vapors would tend to produce true reversals. Discussion of this question must be reserved, however, imtil the spot spectra can be more thoroughly studied with this point in view. In Table IV the mean residual, taken without regard to sign, is 0.021 Angstrdms. If we omit the enhanced lines, because of their exceptional behavior in spots, the mean residual is reduced to 0.015 Angstrdms. As the spot lines range in width from 0.14 to 0.24 Angstrdms, the agreement is closer than would be expected to result from chance alone. When it is remembered that one or more second- ary causes may also affect the width of the lines, the probability that a true relationship exists appears to be considerably increased. A more refrangible region of the iron spectrum gives the results detailed in Table V. 381 20 GEORGE E. HALE Here the mean residual is 0.021 and the range in the width of the spot lines from 0.09 to 0.20 Angstroms. ^5107.82, ^5123.90, X 5198.89, and X 5434. 74, which are very narrow in spots, are not quite separated on the laboratory plates. X 5455 . 83, on the contrary, is single in the laboratory and fairly wide in spots. In this case, at least, there must be some cause of widening in spots other than a magnetic field. A still more refrangible region of the iron spectrum gives the results contained in the following table: TABLE VI Widths of Ison Linxs in Spots AA, Spait WaTc-Length AA.Spuk fl.Z Width in Spots < 4427.48 0.32 0.15 0.16 +0.01 4433.39 0.28 0.13 0.12 —0.01 4443-51 0.35 0.17 0.18 +0.01 444336 O.IO 0.05 0.12 +0.07 4454.55 0.26 0.12 O.IO —0.02 4459.30 0.32 0.15 0.12 -0.03 4461.82 0.32 0.15 0.14 — O.OI 4466.73 0.26 0.12 0.15 +0.03 4469 -54 0.32 0.15 0.12 -0.03 4484.39 0.29 0.14 0.12 —0.02 4494.74 0.25 0.12 0.14 +0.02 4522.80 0.20 O.IO 0.08 —0.02 Enhanced line 4525.31 0.32 0.15 O.II — 0.04 4528.80 0.27 0.13 0.13 0.00 4531.33 0.29 0.14 0.12 —0.02 4548.02 0.22 O.IO O.IO 0.00 4549.64 0.24 O.II O.IO —O.OI Enhanced line 4556 06 0.27 0.13 0.12 —O.OI Enhanced line 4603.13 0.37 0.18 0.14 —0.04 It is interesting to note the progressive decrease toward the violet in the mean width of spot lines and the separation of the correspond- ing doublets in the spark, as shown by the following table. The (weighted) means represent the three groups of lines given in Tables IV, V, and VI. TABLE VII Meu Wave-Length Spot Lines Mean Width Spa*k DonbleU Mean Separation 6330 5267 4495 0.18 0.14 0.13 0.54 0.42 0.29 38a MAGNETIC FIELD IN SUN-SPOTS 2i Although the rate of decrease is more rapid for the spark doublets than for the spot lines, it must be remembered that in the former case the mean separation of the components is given, while the mean width of the spot lines represents the separation of the components plus their width. The width of the components cannot be determined, except in the case of doublets, and therefore the rate of decrease falls off toward the violet, as the width of the spot lines approaches that of the solar lines. The extremely small average shift of the lines in the violet when the Nicol is rotated is in harmony with this view. A group of twelve spot doublets near X 4395, which belong to several different elements and have not yet been photographed in our labora- tory, afford some additional evidence. The mean separations of groups of spot doublets in the red (Tables I, Fe, and II, Ti), green (Table III, Cr)y and violet (those just mentioned) are given in the following table: TABLE VIII MxAN Wavb-Lxngtb Spot Doctlxts Number Mean Separation 6186 5665 4395 9 8 12 0.137 0.145 0.085 Between X 6186 and X 5665 these doublets show no such progressive change as appears in Table VII. AX Preston's law, -7^= const., has been foimd to hold rigorously only for the lines of certain series. It therefore could not be expected to apply with accuracy here, especially as the lines of different elements are included. Nevertheless it is of interest to determine whether the decrease in the separation of these doublets toward the violet proceeds at a similar rate. If we combine the separations for A 6186 and ^ 5665, we have 0.141 for the mean wave-length X 5941. Then 0.141 (S94i) 0.085 (439s) 383 -=4.4X10-0. 22 GEORGE E. HALE The iron doublets, whose mean separations for a field strength of about 15,000 gausses are given in Table VII, yield the following results. 0.44 (SS44)' 0.29 (449s)" i4.3Xio-» 14. 3X10-9. Thus the iron doublets follow the law very closely, while the approxi- mate agreement with the spot doublets, though perhaps the result of chance, is not without interest. Table IX gives the widths of various titanium lines in spots, and the separations of the components of the corresponding doublets, observed abng the lines of force in a field of 12,500 gausses. TABLE IX Widths of Titanixtm Lines AA, Spark WftTe-Length AA, Spark 5-4 < 5823.91 single 0.13 5866.68 0.48 0.14 O.Z9 + 0.05 5880.49 0.64 0.19 0.19 0.00 5899.52 0.50 0.15 0.18 + 0.03 5903-56 0-73 0.3I O.Z9 — 0.03 5918.77 0.73 0.3I 0.30 — O.OI 5932-33 single O.I3 5938.04 0.74 0.33 0.17 — 0.05 5953.39 0.52 0.15 0.13 — 0.03 5966.06 0.47 0.14 o.z6 + 0.03 5978.77 0.38 O.II O.Z4 + 0.03 6064.85 0.88 0.36 0.37 + 0.01 6085.47 0.81 0.34 0.30 — 0.04 6091.40 0.65 0.19 0.15 — 0.04 6093.74 0-55 0.16 0.16 0.00 6098.87 0.59 0.17 0.15 — 0.03 6iai.22 0.56 o.z6 O.I7 + 0.01 6136.44 0.64 0.Z9 0.17 — 0.03 6146.44 single O.I3 6361.33 0.41 O.I3 0.16 + 0.04 6317.67 0.43 O.I3 O.I3 0.00 6336.33 0.56 0.16 0.15 — O.OI 6366.56 0.55 0.16 0.18 •f o.oa For titanium in this region the mean residual is 0.021 Angstrdms for spot lines ranging in width from o. 12 to o. 27 Angstr5ms. 384 T MAGNETIC FIELD IN SUN-SPOTS 23 SIGN OF THE CHARGE THAT PRODUCES THE FIELD IN SUN-SPOTS If the evidence presented in this paper renders probable the existence of a magnetic field in sun-spots, it is of interest to inquire concerning the sign of the charge which, according to our hypothesis, produces the field. In Lorentz's theory of the Zeeman effect in its simplest form, the motion of a single electron in a molecule of a lumi- nous soiurce is discussed.* This electron is supposed to be capable of displacement in all directions from its position of equilibrium, toward which it is drawn by an elastic force, which is proportional to the displacement but independent of its direction. Let e be the charge of the particle, m its mass, fr the elastic force caused by a displacement r, / being a positive constant. The frequency of the vibrations, whether they be linear, elliptical, or circular, will be We may now suppose the light-source to be placed in a homogeneous magnetic field of intensity H. A particle carrying a charge e, and moving with velocity v, will be subjected to a force perpendicular to the field and to the direction of motion of the particle, the magni- tude of which may be represented by evlIsin(Vj H). It is evident that the electron may have three different motions, each with its own frequency. Linear vibrations parallel to the lines of force, having the frequency n^, will not be affected by the magnetic field. Circular vibrations in a plane perpendicular to the lines of force will be affected differently, depending upon whether they are right-handed or left- handed. If r is the radius of a circular orbit and n the frequency, the velocity of the electron will be v=«r and the centripetal force will have the value mn'r. We may now consider the effect on the motion of the electron of the elastic force /r and of an electromagnetic force For a positive charge the latter force is directed toward the center if the motion is clockwise, as seen by an observer toward whom the lines of force are directed. We then have X The following outline of the theory is taken from Lorentz's "Theorie des ph^- nom^nes magn^to-optiques r^emment d^couverts," Rapports, Congr^s international de physique, 3, z, 1900. 385 24 GEORGE E. HALE This frequency n differs very slighdy from the frequency »oi thus the last term of the equation must be much smaller than the term/fi so that we may write am This expression gives the frequency of the right-handed (clockwise) vibrations. For the left-handed vibrations we have eH ,. am As seen along the lines of force a single line in the spectrum is thus transformed into a doublet, the components of which are circularly polarized. An observer toward whom the lines of force are directed will find that the light of the component of greater wave-length, whose frequency has been decreased by the field, is circularly polarized in the right-handed or clockwise direction. Hence (2) is greater than (i), and it follows that the charge e of the electron which produces the spectral lines must be negative. In the case of the solar vortices we have to consider two sets of charged particles, which may be entirely distinct from one another: (i) those whose vibrations give rise to the lines in the spectra of spots, and (2) those that carry the charge which, by the h3rpothesis, produces the magnetic field. The Zeeman effect supplies the means of deter- mining the direction of the lines of force of the sun-spot fields, and photographs of the vortices, made with the spectroheliograph, indicate the direction of revolution of the particles. Thus we are in a position to determine the sign of the charge carried by the particles which pro- duce the fields. As pointed out independently by K5nig and Comu, the violet component of a magnetic doublet observed along the lines of force is formed by circular vibrations, having the direction of the current flowing through the coils of the magnet.* From observations of circularly polarized light, made in our Mount Wilson laboratory by Dr. St. John and confirmed by myself, it appears that when the Nicol prism of the tower spectrograph stands at 60^ E. it transmits the violet component of a doublet produced in a magnetic field directed toward the observer. From Biot and Savart's law the direction of > See Cotton, Le pMnofukne de Zeeman, chap, vii; KOnig, Wied. Ann., 62, 340, 1897. 386 MAGNETIC FIELD IN SUN-SPOTS 25 the ciirrent causing such a field is counter-clockwise, as seen by the observer. In the same position the Nicol also transmits the violet component of a doublet produced in a sun-spot siurounded by a vortex in which the direction of revolution is clockwise. As a nega- tive charge revolving clockwise produces a field of the same polarity as an electric current flowing coimter-clockwise, we may conclude that the magnetic field in spots is caused by the motion of negative corpuscles. PROBABLE SOURCE OF THE NEGATIVE CORPUSCLES We may now consider the probable source of a sufficient number of negative corpuscles to produce a field of about 2900 gausses in sun-spots. In his Conduction of Electricity through Gases^ p. 164, J. J. Thomson writes as follows: We thus are led to the conclusion that from an incandescent metal or glowing piece of carbon "corpusdes" are projected, and though we have as yet no exact measurements for carbon, the rate of emission must, by comparison with the known much smaller rate for platinum, amount in the case of a carbon filament at its highest point of incandescence to a current equal to several amperes per square centimeter of surface. This fact may have an important application to some cosmical phenomena, since, according to the generally received opinion, the photosphere of the sun contains large quantities of glowing carbon; this carbon will emit corpuscles unless the sun by the loss of its corpuscles at an earlier stage has acquired such a large charge of positive electricity that the attraction of this is sufficient to prevent the negatively electrified particles from getting right away from the sun; yet even in this case, if the temperature were from any cause to rise above its average value, corpuscles would stream away from the sun into the surrounding space. On another page (168) Thomson also remarks: " The emission of the negative corpuscles from heated substances is not, I think, confined to the solid state, but is a property of the atom in whatever state of physical aggregation it may occur, including the gaseous." After illustrating this in the case of sodiiun vapor, Thomson adds (p. 168) : The emission of the negatively electrified corpuscles from sodium atoms is conspicuous as it occurs at an exceptionally low temperature; that this emission occurs in other cases although at very much higher temperatures is, I think, shown by the conductivity of very hot gases (or at any rate by that part of it which is not due to ionization occurring at the surface of glowing metals), and especially by the very high velocity possessed by the n^ative ions in the case 387 26 GEORGE £. HALE of these gaaes; the emission of negatively electrified corpuscles from atoms at a very high temperature is thus a property of a very large number of elements, possibly of all. Thus the chromosphere, as well as the photosphere, may be re- garded as copious sources of negatively electrified corpuscles. The part played by these corpuscles in sun-spots cannot be advantage- ously discussed imtil the nature of the vortices is better understood.' At present it is enough to recognize that the supply of negative elec- tricity appears amply sufficient to accoimt for the magnetic fields. Let n be the number of corpuscles per unit cross-section passing a given point in imlt time and e the charge on each corpuscle. Then we have, for the current carried by the corpuscles, c=^ne. H. A. Wilson found that in a vacuum tube, at pressures up to 8 . 5 mm, the ciurent at the cathode was 0.4^ miUiamperes per sq. cm, where p is the pressure in millimeters.* If ^=8.5, we have ^=3 .4X 10"' amperes. Assume the velocity of the corpuscles in this case to be of the order of 10^ km per sec. In a solar vortex (if the charged particles are carried with it) the velocity may be taken as of the order of 100 km per see.* Then if the number of corpuscles per sq. cm were the same in the two cases, the current in the sun would be of the order of 3 . 4X 10"* amperes per sq. cm at the same pressure. We may now assume the corpuscles to be moving at a velocity of 100 km per second in an annulus 25,000 km wide, 1000 km deep, and 100,000 km in diameter surroimding a sim-spot. Taking the current strength to be as above, 3 .4X io~^ amperes per sq. cm, the intensity of the resulting magnetic field comes out 1000 gausses. Such a calculation is of little value, except for the purpose of indi- cating that a magnetic field of the observed order of magnitude might conceivably be produced on the sim.* -EXTERNAL FIELD OF SUN-SPOTS We have already seen that the strength of the field in spots appar- ently changes very rapidly along a solar radius, and is small at the upper level of the chromosphere. < For this reason a discussion of the very interesting suggestion of Professor E. F. Nichols, that the positively and negatively charged particles are separated by cen- trifugal action in the spot vortex, is reserved for a subsequent paper. » PhUosopkical MagoMine (6), 4, 6x3, 2903. s Solar VorHces, p. 13. 4 See a similar calculation by Zeeman in Nahire for August 20, 1908. 3^ MAGNETIC FIELD IN SUN-SPOTS 37 If subsequent work proves this to be the case, it will appear very improbable (as indicated by theory) that terrestrial magnetic storms are caused by the direct e£Fect of the magnetic fields in sun-spots. We have some reason to think that their origin may be sought with more hope of success in the eruptions shown on spectroheliograph plates in the regions surroimding spots. CONCLUSION Although the combined evidence presented in this paper seems to indicate the probable existence of a magnetic field in sim-spots, the weak points of the argument should be clearly recognized. Among these are the following: 1. The failure of our photographs to show the central line of spot triplets before the spots are very dose to the limb. 2. The absence of evidence to support the h)rpothesis that the imperfect agreement between spot and laboratory results is due to differences in the mean level of absorption. 3. The apparent constancy of the field strength, as indicated by the nearly imiform width of the doublets in different spots. 4. The difficulty of explaining, on the basis of our present frag- mentary knowledge of solar vortices, the observed strength of field in the umbra and penumbra, and especially its variation with level. As the resolving power of the 3C>-foot spectrograph is sufficient to resolve completely only the wider spot doublets, the central line could not be separately distinguished in other cases, even if it were present. Hitherto it has been possible to photograph the spectra of only the largest spots, because the images of other spots, as given by the tower telescope, are too small. The need of a telescope giving a much larger image of the sim, and a spectrograph of greater resolving power and focal length, which has been felt in previous work, is strongly emphasized by this investigation. Such apparatus would also permit the spectrum of the chromosphere, and many other solar phenomena, to be studied to great advantage. As regards the nature of the vortices, the principal question is whether the gyratory motion primarily concerned in the production of the magnetic field is outside the boundaries of the spot or within the umbra. In the former case we must face various difficulties, 389 28 GEORGE E, HALE such as the apparent constancy of the field in different spots, and the fact that its intensity rapidly decreases upward. The view that the field is produced by the gyratory motion of vapors within the umbra raises other difficulties, which may also be serious. Fortunately there is reason to hope that observations now in progress may throw light on several of these questions. Mount Wilson Solar Observatory October 7, 1908 ADDENDUM The fact that the doublets in the sun-spot spectriun do not change to triplets, even when the spot is as much as 60*^ from the center of the sun, appeared, when the proof of the above paper was corrected, to be a serious argument against the magnetic field hypothesis. Thanks to the recent work of Dr. King, this difficulty no longer exists, at least in the case of several iron and titanium lines. Photo- graphs of the spark spectrum in a strong magnetic field, taken at right angles to the lines of force, show that the iron lines XX 6213 . 14, 6301.72, and 6337.05 are doublets,* with no trace of a central com- ponent. As these lines are also doublets when observed parallel to the lines of force, it is only natural that they should be double in spots, wherever situated on the solar disk. A 6173.55, which is a fine triplet in spots, is a triplet when observed at right angles to the lines of force. X 6302 . 71 is a triplet, though in Table I it is classed as a spot doublet. In the spot spectrum the line is a triplet, but so decidedly asymmetrical in appearance that I supposed the intermediate line to be due to some element other than iron, greatly strengthened in the spot. It now turns out, however, that the apparent as)an- metry is due to a line of some other substance, which occurs in the spot spectrum. In the laboratory this triplet is symmetrical, though the overlapping component of an adjoining line gave an appearance of asymmetry in the earlier photographs." * Later results prove these to be quadruplets, with the components of each mem- ber of the doublets too close for resolution in the spot spectrum. « The triplet X 6302 . 71 was incorrectly described in this paper, as printed in the Asirophysical Journal for November 1908. The data given above have been derived from later and better photographs. 390 MAGNETIC FIELD IN SUN-SPOTS 29 The titanium lines XX 6303.98 and 6312.46, which are double in spots, are also double in the spark, when observed at right angles to the lines of force.' X 6064.85, already mentioned as a triplet in spots, with a rather faint central component, is a triplet, with strong central component, in the spark under the above conditions. The titanium spot doublets XX 5903 . 56 and 5938 . 04 (Table II) have not yet been observed at right angles to the lines of force. These results leave no doubt in my mind that the doublets and triplets in the sun-spot spectrum are actually due to a magnetic field. As I am now designing a spectrograph of 75 feet (23 m) focal length, for use with a tower telescope of 150 feet (46 m) focal length, I hope it may become possible to investigate small spots, as well as large ones, and to resolve many of the close doublets and triplets in their spectra. > Later results prove these to be quadruplets, with the components of each mem- ber of the doublets too dose for resolution in the spot spectrum. November 4, 1908 END OF VOL. I 391 h X 1^ ^eo SJLo^J^'O (Ewttut^lt 9itfitttiitUiit vi Ulas^ttistott .. i c0ntribut?i0ns ^om the mount wilson solar Observatory NO. 81 1 i on the absorption of light in space BY J. c. kapteyn t' Reprinted from the AUrophysicai J^umai^ Vol. XXIX» January 1909 * V '^.C^, The prcx)f is not complete because different galactic latitudes were not separately treated. I have, however, convinced myself, in a rough way, that thinning-out of the stars is found for every zone of galactic latitude. This, together with the fact that the distribution of the stars varies yery little with the galactic longitude, proves our conclusion. * A beginning has been made in Proceedings of the Academy of Science at Amster- dam, 1908, p. 626. 3 Plan of Selected Areas, p. 57. 2 ABSORPTION OF LIGHT IN SPACE 3 Meanwhile it cannot be denied that the truly fundamental impor- tance of the absorption of light makes it highly desirable to find still other methods, depending on quite different data. Such a new method could probably be obtained if it turns out that the absorp- tion in mterstellar space is more or less selective. If the absorption and scattering of light by meteoric matter is really sensible, then there can be no reasonable doubt but that the violet end of the spectrum must be more strongly affected than the less refrangible rays. But even the probability of gas-absorption, producing absorption lines or bands in the spectrum, cannot be denied a priori. Owing to the gas of the corona lost by the sun, to similar loss presiunably suffered by the other stars, to that lost by comets, etc., interstellar space must contain, at every moment, a considerable amount of gas. Might not this gas, in a thickness of hundreds of Ught-years, cause an appreciable absorption of light? It thus seems that there is reason to inquire whether or not light suffers in space (a) a general absorption, particularly strong in the violet; (i) a selective absorption, widening or strengthening Unes or bands in the spectrum. To my regret I have never had the opportunity of undertaking the observations necessary to settle these questions. Recently, how- ever, I obtained some results from the Harvard observations, which can be explained most readily by an absorption of the first kind. In the observations of the spectra of the bright stars {Harvard Annals, 28, Part I) Miss Maury distinguishes, in Class XVa, two divisions. One of these has the remark 184, the other the remark 185. The parts of these remarks to which I wish to call attention are as follows: " 184. This star resembles a Bootis in the absorption shown in the region having wave-length shorter than 4307." " 185. This star resembles a Cassiopeiae in the general absorption shown in the region having wave-length shorter than 4307." In the detailed description of the classification (p. 39) the differ- ence is described in these words: The stars of this group QCVa) appear to fall into two divisions, exhibiting a slight difference in the degree of general absorption in the violet regions. Of these divisions a Bootis and a Cassiopeiae are respectively typical. In the first 3 4 /. C. KAPTEYN the gcQeral absorption is slight; in the second, it is more conspicuous both in the regions of the violet above mentioned and bejrond wave-length 3889, where the photographic spectrum generally appears to be suddenly cut off. The distinction here made is a very difficult one in practice, owing to the fact that different photographs of the same stars show considerable difference in the general absorption of the violet. The difficulty is so great that Miss Cannon, in the classification of the southern stars, gave up this division altogether. She says {Harvard Annals y 28, Part II, p. 159), "These effects were found to vary so much in different photographs of the same star taken imder different conditions that they have been assumed to be photographic effects, rather than real." On the other hand. Miss Maury contends {ibid.y Part I, p. 39), "The degree of absorption indicated also varies in photographs of different density so that, while there seems to be sufficient ground for believing thai the distinction is a real one, there is more or less lior- bility to error in assigning individual stars to one or the other diinsionJ^ I have taken the liberty of italicizing the words which, in order to appreciate what follows, we must keep well in mind. I have been led to think, with Miss Maiuy, that the distinction is probably a real one and that the most natural, though not the only way to account for it is to attribute it to space-absorption. The suspicion that it might be due to space-absorption was of course at once raised by the preceding considerations. It is easy to see how it may be tested. If it is well founded, then the stars belonging to the a Cassiopeiae division must be farther removed from the solar sjrstem than the stars of the a Bootis division. Our knowledge about individual parallaxes is far too defective to be of any use in the present question. In the absence of the parallaxes the amount of proper motion will be the safest guide in our estimate of the evidence. If our suspicion is in accord with nature, we have to expect that the stars in the a Cassio- peiae division will in general have smaller proper motions than the stars in the division of a Bootis. The following summary shows the proper motions for all the stars in the two divisions. With very few exceptions they were taken from Newcomb's Fundamental Catalogue. Those which are not 4 ABSORPTION OP UGHT IN SPACE contained in that catalogue were taken from Auwers-Bradley as corrected in Graningen Publications, No. 9. All the proper motions may thus be considered as being very reliable. To the stars in Class XVa were added nine stars from Class XIV and XIV for which, according to the remarks of Miss Maury, the tt CASSIOPEIAE GROUP No.mH.P. Scar Kag. in JET. P. aigoo S1900 SpwtTum Cent P.M. 45 iCai 3.6 0^14™ - 9^23' \ 3" 9a 9 Andromedae 3-4 34 +30 19 17 94 a Cassiopeiae 2.2 35 +56 00 6 103 fiCeU 2.Z Z^ -18 32 23 123 62 Piscium 6.0 43 + 645 7 155 43 H Cephei 4.5 55 +85 43 8 183 vCeti 3.6 I 3 -10 43 25 220 a Ceti 3.8 19 - 8 42 ^S 251 V Persei 3.7 32 +48 07 1 13 333 a Arietis 2.0 2 2 +23 CO 1 24 498 K Persei 4.0 3 3 +44 29 1 ^3 560 9 Persei 4-4 24 +47 39 1 3 750 8 Tauri 4.0 4 17 + 17 18 1 II 775 e'TauH 3-9 23 + 15 44 1 7 1452 26 Afonocerotis 4-2 7 36 - 9 19 1 8 1453 a Geminorum 4.1 37 + 29 07 1 23 1629 ^ Hydrae 3-3 8 so + 6 19 I 9 1695 iH Draconis 4.6 9 23 +81 46 \xV.a 3 1760 Ai Leonis 4.1 47 + 26 29 24 1800 X Hydrae 3-9 10 6 -II 51 1 22 1902 46 Leon, min. 3-9 48 +34 45 1 30 2056 14 H* Draconis 5.8 12 + 77 28 1 16 2063 tCorvi 31 5 —22 03 1 7 2208 e Virginis 30 57 + 11 30 1 27 3569 I Draconis 3-3 15 23 +59 19 1 I 2627 a SerpenHs 2.7 39 + 644 I 14 2962 fi Ophiuchi 2.9 17 38 + 4 36 I 16 3003 i Draconis 3-9 52 + 56 55 1 13 3007 ^Herculis 3-9 54 + 29 15 1 9 3084 8 SagiUarii 2.8 18 15 -29 52 1 5 3328 T Draconis 4.5 19 17 + 73 10 1 17 3519 ig Vulpeculae 5-8 20 8 + 26 31 1 3732 tCygni 3.5 21 9 + 29 49 1 6 3836 9 Pegasi 2.4 39 + 9 25 / 2 3923 r Cephei 3-5 22 7 +57 43 / 2 4182 7 Cephei 3-4 23 35 + 77 04 / 17 994 fi Leporis 30 5 24 -20 51 XIV. a 9 1312 ox Canis majoris 4.0 6 50 -24 04 XIV, a 4 1474 i Argus 3-4 7 45 -24 37 XIV. P 1747 9 Leonis 31 9 40 + 24 14 XIV. P 5 2249 y Hydrae Z'3 13 14 -22 39 XIV. a 8 2515 pBootis 3.6 14 58 +40 47 XIV. a 6 3779 i* Capricomi 3.8 21 21 -22 51 XIV. P 2 3780 J5 Capricomi 6.0 21 22 -21 37 XIV. a 6 4020 71 Pegasi 3.1 22 38 +29 42 XIV. a 4 /. C. KAPTEYN a BOOTIS GROUP No. InH.P. Star Mag. a igoo < 1900 Spcctmin Cent. P.M. ■ m 54 Pisdum 6.1 o»»34« + 2o"43' 60" 93 55 Pisdum S-5 35 + 20 54 \ 3 812 jj Eridani 3-9 4 34 -14 30 1 18 1086 8 Leparis 4.0 5 47 -20 54 1 69 1093 9 Aurigae 3.8 51 + 54 17 1 16 I4S9 fi Geminarum Z.I 7 39 + 28 16 1 62 1823 7 Leonis 2.2 10 14 + 20 21 1 34 1926 a Urs, maj. 2.0 58 + 62 18 1 14 1961 w Urs. maj. 3.8 XI 13 +33 39 1 4 ^9^3 BCrateris 3-9 14 -14 14 f 23 1999 2 Draconis 5-5 30 + 69 52 1 16 2400 a Bootis 0.0 14 II + 19 44 \ 228 2541 9Bootis 3-5 15 12 •^33 42 )xv.« 16 3589 y Librae 4.0 30 -14 27 / 7 2736 t Ophiuchi 3-4 16 13 - 4 27 1 9 2774 P Herctdis 2.8 26 + 21 42 I II 3037 70 Ophiuchi 4.1 18 00 + 2 32 1 "5 3090 ij Serpentis 3-4 16 - 2 56 1 90 3447 € Draconis 3-9 19 48 + 70 01 I 9 3450 P Aquilae 4.0 50 + 6 10 1 48 3648 e Cygni 2.7 20 42 +33 36 1 49 3656 iy Cephei 3.6 43 +61 26 1 83 4034 fi Pegasi 3-2 22 45 + 24 05 / 16 4114 y Pisdum 3.8 23 12 + 2 44 1 75 4174 X Andromedae 3 9 33 +45 56 45 absorption for wave-lengths bdow 3889 approaches that of a Cas- siopeiae. From this summary we see that neither are the proper motions of the stars in the a Cassiopeiae division exclusively small, nor the motions of the a Bootis stars invariably considerable. After the words quoted from both Miss Cannon and Miss Maury such a thing could not be expected. Moreover, at least some of the proper motions of the a Bootis stars may only appear to be small because they are seen strongly foreshortened, and at least a few of the proper motions of the a Cassiopeiae stars must be large, not owing to small distance, but in consequence of an exceptionally large linear velocity. Finally, in classifying the stars into two divisions gradually merging into one another, there must always be a certain proportion so near the Une of division that they might almost equally well be put in either class. In this respect it is worthy of notice that for Class XIV Miss Maury made no division on account of absorption in the violet. For this class we may thus presume that she will have drawn attention only 6 ABSORPTION OP UGHT IN SPACE to the well-marked cases. For this reason the fact that in all these cases, without exception, the proper motion is small seems very sig- nificant Notwithstanding the contrary cases, our list shows clearly that, as a ride, the proper motions in the a Bootis division exceed those in the other group very considerably. The following averages may show this yet more forcibly. General mean of centennial motion !a— oh to a « 6** ^ 1° " 12 to l8 i8 to 24 Number of stars with centennial p.m. < 10" Number of stars with centennial p.m. >3o" a CasnofetoB div. iif4 (45 stars) 13.6 (15 stars) 12.9 (10 stars) 1 1. 9 (xo stars) 6.1 (10 stars) 58 per cent, o per cent. a Bootis diy. 47 f I (25 stars) 33.0 ( 5 stars) 25.4 ( 6 stars) 54.1 ( 6 stars) 58.9 ( 9 stars) 20 per cent. 48 per cent. The percentages become still more favorable to our supposition if we include only the cases in which two or more, or three or more plates have been taken. As the number of stars becomes rather small in these cases I will not further insist on this point. The objection which might be based on the fact that the subdi- vision here discussed was made only in the case of Class XVa, whereas, if the interpretation by space-absorption is correct, it must show in all classes, loses greatly in force by the fact that all other classes of Miss Maury contain so small a number of stars that the contrast between the two divisions had small chance of being noticed. If we grant the reality of the difiFerence in the distance of the stars of the a Bootis and a Cassiopeiae divisions, it does not neces- sarily follow that the interpretation of the difference in the spectrum as being due to space-absorption must be correct. For, as the average magnitude of the stars in the two divisions is about the same, the total light-power (luminosity) of the more remote a Cassiopeiae stars must be greater than that of the a Bootis stars. The meaning of our result may therefore also be: for stars of the same spectrum, the light of the more luminous objects is relatively weaker in the violet part of the spectrum. Whichever view we think the more probable, the phenomenon seems well worthy of being more closely and carefully investigated. Of course care will be taken to make the spectra to be compared 7 8 /. C. KAPTEYN as equal as possible in the less refrangible part. Moreover, we shall choose exclusively extreme cases in regard to proper motion. In particular, fainter stars ought to be included in order that we may have a more extensive choice of stars having large proper motions and at the same time a larger difference of distance of the two groups to be compared. It seems unnecessary to enter into further details on this point. But a few lines may perhaps be allowable in order to show the importance of a determination of the space-absorption which might eventually result from such an investigation. There is first, of course, the contribution we may expect for the determination of the tokil amount of absorption (the absorption- constatU in the case the imiverse is fairly imiformly filled with the absorbing matter). As was pointed out above, such a determina- tion would probably enable us to get rid of the main obstacle in the way of a reliable determination of the real distribution of stars in space. Then there is the fact that we shall get a criterion or even a measure of distance, which promises to be of particular value where other means begin to fail. To see this clearly we have only to con- sider how uncertain our determinations of average parallaxes, even if obtained by means of the parallactic motion, become for stars for which the distance exceeds, say, 3000 light-years. In the present state of scieilce the discrimination between a distance of 3000 and a distance of 6000 light-years must be considered as practically impos- sible. On the contrary, if space is somewhat imiformly filled with matter, the difference between the absorption for the two classes of stars must be equal to that between stars at the respective distance zero and 3000 light-years. Further, there is the importance of getting an insight into the true spectrum of the stars, freed from the changes brought about by the medium traversed by light on its way to the observer. Thus, for instance, is it not possible that the somewhat strange phenomenon, that, according to Huggins, stars of Secchi's second type often show spectra which, in the more refrangible part, are more intense and more extensive than in the spectra of the first type, is simply due to space-absorption? For it is well known that the average distance of the stars of the second iypt is considerably smaller than that of the stars of the first type of equal magnitude. 8 ABSORPTION OF UGHT IN SPACE 9 Even more important than the general absorption here discussed would be a gas-absorption, producing space-lines or space-bands. Indications of such an absorption might be obtained as above by comparing stars having spectra of the same class but of widely different proper motion. There would be, however, a decisive criterion by which we might prove or disprove the reality of our results. If there are space-lines, they must not share in that part of the radial motion which is due to the motion of the stars themselves. The space- lines must thus show their nature in somewhat the same way as the atmospheric lines of the solar spectrum in Comu's well-known experiment. As, however, I have no evidence as to the real occurrence of such lines or bands, no more need be said about them at present. I am permitted to say in conclusion, that in the program of the stellar work for the five-foot reflector of the Mount Wilson Solar Observatory some stars chosen with special reference to the subject of the present paper will be included. Mount Wilson Solar Observatory October 1908 h. < ■ ) "X f k y ^•cJ S^0.36'0 r it * - X (Santrgir SttstitittUm itf 9laai|ttigtmt Contributions from the Mount Wilson Solar Observatory NO. 32 V L- THE RELATIVE INTENSITIES OP THE CALCIUM LINES H, K, AND X 4^6 IN THE ELECTRIC FURNACE Bt ARTHUR S. KING r; r I Reprinted from the Astrophysical Jdumaly Vol. XXVIII, December 1908 ■■A I i (i.^u.9^ e V ♦ Contributions from the Mount Wilson Solar Observatory, No. 32 Reprinted from the Asirophytical Jowmci^ Vol. XXVIII, pp. 389-396, xgoS THE RELATIVE INTENSITIES OF THE CALCIUM LINES H, K AND A 4227 IN THE ELECTRIC FURNACE By ARTHUR S. KING A number of investigations have dealt with the changes in relative intensity of the blue and violet lines of calcium according to the manner of producing the spectrum. Interest has centered about the behavior of X 3969 and A 3934 (H and K in the solar spectrum) as compared to that of X 4227 {g in the solar spectrum) ; the appearance of these lines having been noted by a number of observers in the flame, arc and spark and by the writer in the electric furnace. Some photographs obtained with the large electric furnace in this laboratory have furnished additional data regarding the dependence of these lines on the physical conditions in the light-source. These results will now be given, with an attempt to summarize the present condi- tion of our knowledge based on the action of the lines in various light-sources. The line A 4227 has always been. recognized as a "flame line," appearing easily in the flame and in all laboratory sources where a considerable amoimt of calcium is present. It was shown by Sir William and Lady Huggins' that when a very minute trace of calcium is present on platinum or iron terminals in the electric spark the line A 4227 does not appear (for photographic exposures of ordinary length), while H and K are yet distinctly visible. The other extreme is foiind in the low-temperature flames, where A 4227 alone is visible and may be obtained very strong if a large amount of calcium is supplied to the flame. In the oxyhydrogen and other very hot flames the H and K lines appear in the core of the flame, but are weak in com- parison to X 4227 and become weaker as we pass away from the cen- tral portion, the outermost part of the flame giving A 4227 alone. The electric arc gives a similar gradation with higher temperature conditions, the arc at moderate current giving H and K much stronger than X 4227 in the core of the arc, the relative intensity changing as > Astrophysical Journal, 6, 77, 1897. 11] I 2 ARTHUR S. KING we go outward until in the "flame" of the arc A 4227 is much the stronger, although H and K are still distinct. The electric spark with self-induction, being near the arc conditions,* shows A 4227 stronger than H and K; while in the highly condensed spark the latter lines are much the more intense. In all these cases a plentiful supply of calcium is assumed to be present. The amount of calcium vapor in the light-soiu'ce has considerable influence, A 4227 in the arc becoming wide or narrow as the supply of calcium is greater or less much more rapidly than do H and K; while in the spark the experiment of Huggins was gradually to diminish the amoimt of calcium on the electrodes until H and K alone remained. This effect is complicated in the arc and spark, however, by the change in discharge conditions given by varying the amount of metallic vapor. The effects thus far reviewed indicate that the intensities of H and K are governed largely by the temperature of the source, the con- densed spark being assumed to give a condition equivalent to that of very high temperature. A comparison of the flame spectrum of cal- cium with the electric-furnace spectrum obtained by the writer in a former investigation ^ indicated, however, that other conditions in the flame may especially favor the H and K lines and exert more or less influence on their production in the arc and spark Metallic calcium was vaporized in a graphite resistance tube in an atmosphere of pure hydrogen. The H and K lines did not appear, though most of the prominent calcium lines were present. H and K appeared, however, from traces of calcium salts occurring as impurities when other substances were vaporized in the tube. The temperature of this furnace was not measured, but there is little doubt that it was as high as that of some of the flames in which H and K are given strongly. This pointed to the conclusion that not only the temperature but the chemical processes of the flame are concerned in the production of these lines. Evidence along the same line is given by a recent work of Hemsalech and de Watteville,* in which they found that with the electric method of volatilization H and K were extremely faint in the flames of air with coal-gas, also of > Astrophysical Journal, 27, 353, 1908. » CompUs Rendus, 147, 188, 1908. 13 11 INTENSITIES OF CA LINES IN FURNACE 3 air with hydrogen, but as soon as oxygen was substituted for air these became the strongest lines of the spectrum with the exception of A 4227. Passing to the results obtained with the new electric furnace in the Pasadena laboratory, described in a previous paper,' Plate I shows a series of calcium spectra given by the furnace imder various conditions. These were photographed with the large Littrow spec- trograph,' using the first order of a plane grating 5 inches (13 cm) long, with 14,438 lines to the inch (567 lines to the mm). The fur- nace was operated in vacuum. The resistance tubes were of Acheson graphite of ^ inch (12 mm) inside diameter, and the alternating current was usually used at 20 and 30 volts. The experiments were made before a Wanner pyrometer was added to the laboratory equip- ment, and as this lot of graphite tubes was exhausted before the temperatmre measiu'ements began, only estimates based on tem- peratures measured in other tubes and on the appearance of the spectra can be given, these tubes being of higher resistance than those for which measurements were given in the previous paper. As nearly as can be judged, the furnace at 20 volts gave 2400^ to 2500^ C, and at 30 volts about 2900^ C. The main purpose of this paper is to show the effect of change of temperatiu^e when different quantities of calcium vapor were present and the temperature intervals given by 20 and 30 volts were large enough to leave no doubt of this effect. Plate I reproduces (i) the spectrum of calcium in the carbon ^^9 (^)i (3)9 ^^^ (4) ^6 spectra given by a small quantity of calcium in the furnace at different temperatmres. Two or three small fragments of clean metallic calcium, weighing together about 0.05 gram, were placed in the graphite tube. As was to be expected, A 4227 was easily obtained, but a high temperature was required to bring out H and K, probably higher than was attained by the writer's former experiments with the furnace in hydrogen when H and K did not appear. Photograph No. 2 shows H and K barely visible and was obtained in a somewhat indirect manner. > CorUribuHons from the Mount Wilson Solar Observatory, No. a8; Astrophysical Journal, 28, 300, Z908. 3 Contributions from the Mount Wilson Solar Observatory No. 27; Astrophysica Journal, 28, 344, Z90S.' 13 4 ARTHUR S, KING The fumade had been used for a series of photographs at 20 and 30 volts lastmg some ten minutes, with the result that the tube and jacketing were very hot. Fifteen volts were then put on the tube and No. 2 taken in ten ndnutes, where one-half to two minutes had been sufficient for strong photographs at the higher voltages. This spec- trum should be compared with No. 3, which is an excellent photo- graph taken with 20 volts on the tube and a temperature that can be safely estimated as 100^ to 200*^ higher than No. 2. This gives the stage when H and K first appear distinctly. It will be noted that X 4227 is fairly narrow and of nearly the same intensity in these two photographs^ while a great change has taken place in the strength of H and K. A third stage in temperature, still with a relatively small amount of calcium vapor present, is represented by No. 4, made with the fur- nace at 30 volts. The negative was somewhat lacking in density, but the character of the lines is distinct. H and K appear as moder- ately strong bright lines with a short exposure. Their width is a little greater than in No. 3. X 4227 is slightly broadened, with a very narrow reversal (not shown in the reproduction). Other photographs taken at the 30-volt temperature were less favorable for reproduction on account of a continuous ground which masked the H and K lines to some extent. This was due presumably to light from the intensely bright walls of the tube being reflected by the particles of vapor. All direct light from the walls was excluded by the method of focusing the interior of the tube on the slit, leaving out the bright image of the walls. The width of H and K and the reversal of X 4227 were prac- tically imchanged, however, although in one photograph the con- tinuous ground was strong enough to give all of these as absorption lines. A reversed line in the furnace spectrum means that the cooler vapor near the ends of the tube is emitting with sufficient strength to give a narrow absorption line through the broader emission line given by the vapor in the hottest part of the tube. The occurrence of true absorption lines would then result when the suspended par- ticles reflected enough light from the incandescent walls, in addition to the continuous spectrum given by the particles themselves as white- hot bodies, to give a background for the entire discontinuous emission of the vapor, thus producing dark lines. 14 INTENSITIES OF CA UNES IN FURNACE 5 Nos. 5 and 6 were taken with the furnace at approximately the same temperature as Nos. 3 and 4 respectively, but with two to three grams of calcium metal in the tube. Therefore some fifty times as much vapor was present as for Nos. 3 and 4. Comparing No. 5 with No. 3, we find H and K slightly broader with more vapor present, but no sign of reversal, thus showing that only the vapor in the middle portion of the tube is hot enough to give these lines. The condition with X 4227 is very dififerent. It appears as a broadly reversed line, the dark portion being of about the same width as the whole bright line of No. 3, showing that the vapor near the ends is both hot and dense enough to emit (and therefore absorb) strongly, while the large amoimt of calcium present causes the vapor in the hottest portion to emit a very broad bright line. Spectnmi No. 6 was taken with 30 volts on the furnace and a large supply of calcium in the tube. A comparison with No. 4, taken at about the same temperature, shows the effect of the increased amount of vapor. In the first place, the continuous groimd is now strong enough to give all of the metallic lines in absorption. H and K are still to be rated as narrow lines. It is difficult to say how closely their emission is confined to the hottest portion of the tube; but at any rate the large amoimt of vapor at this temperature is not sufficient to give any decided broadening for these lines. It is A 4227 which shows an enormous sensitiveness to the increased amoimt of vapor. The width of the whole absorption line must now be compared with the width of the whole emission line of No. 5, not with the width of the reversal in No. 5. The width of A 4227 is thus seen to be only slightly greater in No. 6 than in No. 5; the difference being very nearly the same as for Nos. 3 and 4, which had about the same temperature interval with less vapor present. The slightly increased width in Nos. 4 and 6 as compared to their companion photographs may easily result from the increased vaporization in the tube at the higher tem- peratiu-e. Photograph No. 7 is not a true fmnace spectrum, but it throws additional light on the furnace results and besides was taken under conditions so imusual that it is valuable in itself as presenting a dis- tinct condition of the calcium spectrum. The furnace tube, having been operated for some time at 20 and 30 volts with a strong vapori- 15 6 ARTHUR S. KING zation of the carboiii wore thin and burnt through near its middk. An arc then formed at the break, a heavy current passing at 30 volts, as the large area of the ends combined with the presence of calcium vapor sufficed to maintain the arc at the low voltage. An exposure of 15 seconds then gave photograph No. 7. We thus have a spectrum given by the light from a high-ciurent arc at low voltage passing through some five inches of calcium vapor heated probably well above 2500^ C. There is considerable continuous ground, but H and E show as broad, bright lines, much wider than in No. i (given by calcium chloride in the ordinary carbon arc) and enormously stronger than in any of the furnace spectra. The fact that there is no dark line through the center indicates that the cooler vapor at the ends of the tube does not emit these lines with any appreciable strength. This end vapor may have suffered a considerable fall in temperature owing to the voltage being now largely concentrated at the arc X 4227 is a dark line of about the width of the reversal in No. 5, but without the wide emission line on which the dark line in No. 5 is superposed. The negative shows bright edges to the absorption line, but the arc formed by the broken tube evidently does not give X 4227 with much intensity. We have here, ui fact, very strong arc condi- tions, as indicated by the strength of H and K compared to X 4227: conditions similar to those given by the core of a heavy cinrent arc in air, and more pronoimced than those of the arc at moderate cur- rent shown in No. i. The calcium spectrum as far as the series triplet X 4426-A 4456 is reproduced in Plate I. Numerous relative differences among the calcium lines may be seen at different furnace temperatures and between furnace and arc, notably in the group X 4283-A 4319. These will be considered in a later paper in connection with other regions of the calcium spectrum. DISCUSSION The furnace results show very distinctly the conditions governing the intensity of X 4227. It appears at a low temperature, and is not strengthened in proportion as the temperatiu'e rises. It is, however, very sensitive to changes in the amoimt of luminous vapor and may be enormously strengthened even at moderate temperatures by a large supply of calciimt A comparison of photographs Nos. 3 and 5, 16 INTENSITIES OP CA UNES IN FURNACE 7 taken at approximately the same temperature, brings out this relation very clearly. The readiness of the line to reverse in the arc is due to the strength with which the cooler outer layers of the arc radiate this line and therefore absorb the light from the hotter region of the core. The conditions for reversal are still more favorable in the furnace on accoimt of the length of the radiating column, the cooler ends of which give strong absorption. The production of H and K is more complex, but the furnace shows that they may be obtained if the temperature is sufficiently high (close to 2500^ C.) without the aid of chemical action, although the latter may aid in their production in sources of lower temperature. They are to be rated as high-temperature lines, by reason both of the temperature required for their production and because their increase of strength in the furnace is closely proportional to the tem- perature — ^very much more so than in the case of A 4227. The furnace gave them always as narrow lines, bright imless the continuous back- ground was strong enough to reverse the whole spectrum. Their refusal to show "self -reversal" in the furnace is in line with the fact that their reversal is usually narrow in the arc: the cooler regions in both sources can emit the lines only weakly if at alL It cannot be said, however, that the amount of vapor has no eflFect. The lines widened slightly in the furnace with increase of vapor, but the change is smalL The intensities of the lines in different electrical sources, as the core and flame of the arc, the spark with and without self-induction, are all in line with the hypothesis that the strength of the H and K lines depends very largely upon the temperature of the source. The results from the furnace work do not explain all of the phenomena shown by H and K in the arc. Their intensity in the arc is so much greater that the conditions for their emission seem to be on a different plane from those prevailing in the furnace, the difference being out of proportion to the probable thermal interval between the two sources. Applying these conclusions to solar phenomena, the behavior of the lines H, K and A 4227 in sun-spot spectra does not throw much light on the physical condition of sun-spots on account of the short wave-length of these lines, this region not showing changes in its lines consistent with those observed in the region of greater wave- 17 8 ARTHUR S. KING length. Thus the fact that A 4227 shows little change in sun-spots does not justify statements concerning the vapor-density in spots based on this line. The behavior of the lines in the solar chromosphere is, however, of high interest. Numerous eclipse observations have shown that H and K appear alone in the higher regions of prominences, A 4227 appearing only when the vapor in the chromosphere has reached a considerable density. This is entirely consistent with the conclusions of this paper regarding the conditions favoring these lines. Mount Wn.soN Solak Observatory October 1908 is : r < « .'I V -1^ ^c^ SJLo.^SO <(.MItM of Matftfiitstfiti iontributions prom the mount wilson solar Observatory NO. 88 spectroscopic investigations op the rotation OP the sun during the year 1908 WALTER S. ADAMS •fi Rcpri&ted from tlie ArtrophysUal /wmal. Vol. XXIX, March 1909 i. j»*> ^ * -^ ■■fci^MHi »1.-'. -— - - - \ •J i t > ^*4i^^>V^ '■ I / tii i Contributions from the Mount Wilson Solar Observatory, No. 33 Reprinted from the Ashophyiical Journal, Vol. XXIX, pp. 110-145, 1909 SPECTROSCOPIC INVESTIGATIONS OF THE ROTATION OF THE SUN DURING THE YEAR 1908 By WALTER S. ADAMS The results of a determination of the rotation of the sun based on photographic observations of the displacements of the spectrum Unes were published by the writer in November 1907.* Two conclusions derived from this investigation were of special importance as bearing on a continuation of the work. The first was that a considerably longer series of observations was necessary in order to furnish reliable evidence in regard to a variation in the rate of the solar rotation. The second was that the rotation rate differed for different lines, making it particularly desirable to include in later determinations lines of elements showing a wide range in the altitude which they attain in the solar atmosphere. The observations made during the present year, accordingly, divide themselves naturally into two parts. The first is a direct continuation of the earlier work on the rotation of the general reversing layer with a view to detecting a possible variation in the rate of rotation during the interval covered. The second is a study of the lines of certain elements which are known from in- vestigations of the chromospheric spectrum to rise to great altitudes in the sun's atmosphere. Preliminary values for one of these lines, the a line of hydrogen, have already been published." In the present paper is given a more detailed accoimt of the results of the research. The photographs obtained during 1906-1907 were made with the Snow telescope and the 18-foot (5.5 m) Littrow spectrograph regularly employed with it for general spectroscopic work. During the autumn of 1907 the tower telescope was completed, and the decidedly su- perior advantages which it offered for an investigation of this sort led to its use for all observations made after January 1908. A » Contributions from the Mount Wilson Solar Observatory^ No. 20; Astropkysical Journal^ 26, 203, 1907. > Contributions from the Mount Wilson Solar Observatory, No. 24; Astrophysica- Journal^ 27, 213, 1908. 19] I 2 WALTER 5. ADAMS detailed description of the tower telescope and the 30-foot (9.1 m) Littrow spectrograph used with it will be found elsewhere." The images of the sun formed by the tower are as a rule considerably superior to those given by the Snow telescope. Not only is the defi- nition of the image better, but there is much greater freedom from changes of focus, and the efifect of astigmatism is largely eliminated. The last feature is of especial importance in observations of the rotation, since the efifect of astigmatism is to bring upon the slit light from difiPerent latitudes on the sun's surface, and consequently to introduce systematic errors into the results. As compared with the Snow telescope the tower has but one serious disadvantage, namely, the color-curve of the objective which is used in place of the con- cave mirror of the former instrument. With a knowledge of the form of this curve, however, it is merely necessary to make a corre- sponding allowance in focusing the sun's image upon the slit of the spectrograph, and this has always been done when working in the violet region of the spectrum where the color-curve is comparatively steep. The difiPerence in the size of the visual and the photographic images upon the slit is then readily computed from this difference of focus. The essential difference between the 30-foot spectrograph and the 18-foot instrument of the Snow telescope is that the former is capable of rotation about a vertical axis. This makes it possible to secure any desired position angle on the sim's siurface directly. The spec- trograph is provided with a divided circle 30 inches (0.76 m) in dia- meter upon which the readings of the position angles are made. The spectrograph lens can be focused and the grating rotated by handles which pass up the tube from below, and the frame carrying the plate- holder is arranged so that any required tilt may be given to the plate. The diagonal prism attachment used to bring the opposite limbs of the sun's image upon the slit is considerably more simple than that employed in the first series of observations, owing to the fact that it is fastened directly to the upper end of the spectrograph, and rotates with it, instead of being itself capable of rotation. A small casting about 8 inches high rests on the main plate of the spectrograph, its > Hale, Contributions from the Mount Wilson Solar Observatory^ No. 33; Astro- physical Journal, 27, 204, 1908. 30 ROTATION OF SUN . 3 lower surface coming a short distance above the slit. Four small diagonal prisms are moimted on brass blocks fastened to the imder surface of this casting, two immediately above the slit, and the other two below small slots in the casting. The distance between the centers of these slots is equal to the mean diameter of the sun's image. The second pair of prisms are capable of adjustment toward or away from each other, to allow for the variations in the sim's diameter. Each prism is held in a small moimting and provided with adjusting screws. On the top of the casting is an aluminum plate ruled with several concentric circles by means of which the s\m's image is centered upon the slit. As soon as the prisms had been accurately adjusted so that the grating was uniformly illuminated, the casting was fastened to the top of the spectrograph, its position being accurately defined by tapered pins. It has been found unnecessary to disturb this adjustment in any way, since it is possible to clean the surfaces of the prisms without removing them from their mountings. It is evident that in a form of instrument in which the diagonal prisms are fixed the danger of errors arising from imequal illimiina- tion of the grating surface is much less than in cases where the prisms themselves are movable. I have been careful to test this adjustment frequently, however, usually inunediately before the exposure and again at the end when the spectrograph is rotated 90^ from its original position. The ratio of aperture to focal length in the case of the objective of the tower telescope is i to 60. In a spectrograph of 30 feet focal length, accordingly, a 6-inch grating would be fully illuminated. The one actually used has a ruled surface only 3.25 inches long, and in the higher orders is inclined at such an angle that a beam less than 3 inches in diameter is sufficient to fill it. The factor of safety evidently is considerable. The grating used in this spectrograph is the same as that em- ployed in the previous determination, with the 18-foot instrument, and described in the earlier paper. The photographs in the violet region of the spectrum have all been made in the third order of the grating; those of the a line of hydrogen in the second order. Both of these orders are exceptionally bright, and the definition is excellent. The h'near scale of the plates in the violet of the third order is I mm =0.56 Angstrdm. 21 4 WALTER S. ADAMS The settings of the position circle of the spectrograph correspond- ing to the heliographic latitudes desired are made very readily with the aid of a short table of position angles of the sun's axis. In order to apply these, however, a reference line is necessary, and this has been found by observing the transits of the sun's image across the position circle when the coelostat mirror is rotated. The transits of both limbs are observed and the mean value taken as the transit of the sun's center. The readings on the opposite sides of the circle should, of course, differ by i8o^ if the image has originally been centered upon the slit. The line joining these points gives the true east and west direction, and the calculation of the desired settings becomes very simple. There is a considerable gain of time in obtain- ing the direction of the reference line in this way, rather than by allowing the image of the sun to drift across the position circle, and the adjustment of the instrument has been found to be so accurate that no appreciable error is introduced. In selecting the heliographic latitudes to be observed I have followed the same course as in the earlier series of observations. For the reversing layer an exposure is made at every 15° of latitude be- tween o^ and 90^, and so far as possible for at least one intermediate point in each zone. This gives a total of some twelve to fourteen points from which to determine the velocity curve. In the case of the other Unes the observations have been limited to every 15° of latitude. The addition of the exposure at the pole of the sun furnishes a most valuable check upon the condition of the instrument, or any possible disturbance of the slit which may lead to a change in the inclination of the spectrum lines. This is an important advantage which the present series of plates possesses over the previous one made with the 18-foot spectrograph. With the latter instrument it was rarely possible to reach the sun's pole on account of the interference of the small diagonal prisms which brought the light to the slit. It is unnecessary to describe in detail the method of reduction of the plates, since this differs but little from that given in the earlier paper. The calculations of the latitudes have as before been made with the use of De La Rue's reduction tables, and the quantities to be applied to correct the observed velocities for the departure of the Sim's pole from its visible edge, as well as for the earth's motion, have 32 ROTATION OF SUN 5 been taken from Dim^r's recent valuable memoir.* Most of the reversing layer plates have been measured by Miss Lasby upon the 150 mm Toepfer measuring machine. In the case of the other plates a small Gaertner comparator has also been used. The periodic errors for both instruments are well below the errors of measure- ment. The discussion of the results obtained naturally divides itself into two parts. The first deals with the general reversing layer, by which is imderstood the region in the solar atmosphere at which the absorp- tion takes place that gives rise to the great majority of the narrow lines in the spectrum. The second part deals with those lines upon which special investigations have been made. RESULTS FOR THE REVERSING LAYER In order to facilitate direct comparison with the values obtained during 1906-1907 it is clear that the list of lines upon which the deter- mination is based should be nearly identical with the former list. There are, however, a few lines in the region of the spectrum A 4190 to X 4300, not previously included, to which special interest is at- tached. These have been added to the list, and at the same time two lines in the former list have been omitted. These gave values close to the mean derived from all the lines, and seemed to possess no particular significance. The revised list of lines is as appears in Table I. In Table I four lines have been added to the list used in the determinations of 1906-1907. Of these A 4207, like A 4197 and A 4216, belongs to the violet cyanogen fluting. Of the remaining three lines, A 4283 and A 4289 are due to calcium, and are included for pur- poses of comparison with A 4227, while the line A 4233 is added be- cause of its interesting behavior in the spectrum of the chromosphere. Though assigned in Rowland's table to Mn it coincides with a strongly "enhanced" line of Fe^ and its strength in the chromosphere makes the latter identification the more probable. The series of plates used in this investigation began in February and continued through October, numbering 33 in all. With the exception of July, at least one plate has been taken during each month, « Nova Acta Regiae Socieiatis Scientiarum Upsaliensis^ Ser. 4, Vol. I, No. 6 23 A I WALTER S. ADAMS TABLE I 4196.699 4197-257 4*03.730 4307.566 4316.136 4320.509 4332 •8S7 4233-328 4257-815 4258.477 4265.418 4266.081 4268.915 4276.836 4283.169 4284.838 4287.566 4288.310 4289.525 4290.377 4290.543 4291.630 Element Intensity La 2 C 2 Cr 2 C iN C z Fe 3 Ft 3 Mn 4 Mn 3 Fe 3 Fe 3 Mn 3 Fe 3 -Zf 3 Ca 4 Ni I Ti I Ti, Fe I Ca 4 Ti 3 Fe I Fe 3 Behavior at Limb Much weakened Slightly weakened Strengthened and widened Weakened Weakened Slightly strengthened and widened Much strengthened and widened Much weakened. This is probably not If n, but '^enhanced" line of Fe Slightly streng^ened and widened Much strengthened and widened Slightly weakened Slightly weakened Slightly weakened Weakened Strengthened and widened Slightly weakened Slightly strengthened and widened Widened Probably slightly strengthened Slightly weakened. "Enhanced** line of Ti Slightly weakened Much strengthened SO that the interval may be said to be covered reasonably well. On account, however, of the great variation in the value of the angle the secant of which enters as a factor in correcting the observed velocities for the tilt of the sun's pole, it has been necessary to limit the observa- tions at the highest latitudes to the time when this quantity was smallest. For this reason the plates giving the mean latitude 79? 2 cover but a short period of time. In the case of some of the inter- mediate latitudes, as well, the number of observations has been in- creased toward the end of the series in order to bring the weights of these points into substantial agreement with the others. The ob- servations at 0° and multiples of 15° from that point have been dis- tributed nearly uniformly throughout the series. In Table II is given a summary of the individual plates in the same form as that used in the previous determination. The velocities are corrected for the earth's motion, and are derived from a mean for all the lines. The mean values derived from Table II and grouped about thirteen points of latitude will be found in Table V. Before entering upon a discussion of the results, however, it is desirable to 24 ROTATION OF SUN TABLE II Number of Plate Date 103 1908 Feb. 16 105 1 Mar. 10 « I05a w 106 Mar. 10 Mar. 10 « 113 Aprils 1171 May 26 « 117a May 26 «# I20z June 2 » I20a June 2 Number ^ Number of Lizkes V of Plate km 23 0?2 2.072 w I20a 14.7 1.948 29.6 1-639 44.4 1.289 w 128 59-1 0.776 73-4 0.434 22 03 2.068 14.6 1.982 29.4 1.668 44.2 1-343 « 132 59.3 0.844 74.3 0.408 22 0.4 2.073 15 a 1.960 30.1 1-673 451 1-305 •134 22 60.5 0.794 75 -o 0.416 0.4 2.076 IS-2 1. 971 301 1.659 45 I 1-304 60.0 0.799 «I35i 75 0.408 22 0.0 2-095 14.9 1.967 29.8 Z.682 44.8 1.294 60.7 0.800 75-7 0.363 «I35a 22 0.6 2.056 14.4 1.900 29.4 1.664 44-6 1. 301 60.4 0.827 75-9 0.383 22 0.6 2.060 « 136 14.4 1.902 29.4 1.664 44.6 1.297 60.4 0.845 75.9 0.388 22 ".5 1-965 2.5 2.052 w 146 17.5 1.866 32.8 1.654 48.3 1. 172 63.8 0.713 77-3 0.335 22 10.5 2.015 «I47 4-5 2.039 195 1.842 34.8 1.609 Date Number of Lines ^ X908 June 2 22 5o?3 65 .8 79 3 June 9 22 5 14. 5 29 5 44 5 59 5 74 5 June 10 22 4 4 19 4 34 4 49 4 64 4 79 .4 June II 22 ■5 4 5 19 5 34 5 49 5 64 5 79 6 June II 22 5 14 5 29. 5 44 •5 59 5 74 5 79 5 June II 22 5 14. 5 29 5 44 5 59 5 74 5 79 5 June II 22 5 4 5 19 5 34 5 49 5 64 5 79 5 Aug. 5 22 5 14 6 29 .6 44 9 60 .1 74 9 Aug. 5 22 ■3 14 .6 29 .6 44 9 km 1. 160 0.697 0.254 2.049 1. 917 1.684 1.258 0.829 0.379 2.032 1.896 1.596 1. 169 0-795 0.262 2.048 2.013 1-759 1.562 1. 166 0.707 0.262 2.060 1.940 1.675 1.245 0.773 0.381 0.263 2.038 1-943 1.680 1.238 0.772 0.387 0.266 2.047 1.982 1.848 1-551 1. 149 0657 0.254 2.034 1.934 1.658 1.266 0755 0.403 2.054 1-959 1.687 1.269 25 8 WALTER 5. ADAMS TABLE II— Continued Number Date Number ♦ Number Date Number * olPUte of Lines V of Plate of Lines V xgoS km X908 km « 147 Aug. 5 33 60? I 74-9 0.783 0.403 « 165 Aug. 37 33 49?4 64.9 1. 141 0.694 ta 148 Aug. 5 33 0-3 14.6 39.6 44-9 60.1 74.9 3.056 1-959 1.688 1.365 0.805 0.405 ta 166 Aug. 37 33 4.2 II. 6 19. 1 34.1 49-4 64.9 3.048 1.985 1.875 1.560 1. 149 0.697 M 151 Aug. 6 33 0.1 0.1 3.066 3.059 ta 179 Sept. 30 33 60.1 60.1 0.863 0.865 14.6 I 952 ta 180 Sept. 30 33 60.1 0.803 44-6 1.370 60.1 0.805 59.8 0.8l3 ta 183 Oct. 9 33 II, 1.990 74-9 0.416 II. 1. 991 ta 161 Aug. 36 33 4.1 10.8 3.044 1.989 19.0 33-9 1.843 1.483 19.3 1.886 ta 183 Oct.9 33 33 9 I 548 34.2 1-547 19.0 1.864 49-4 1. 081 II. 1.992 65.0 0,631 II. 1.998 ta 163 Aug. 36 33 4.1 10.8 19.3 34.2 3.048 1.988 1.880 I 504 19.0 33-9 33-9 19.0 1.858 1.549 1-545 1.859 49 4 1. 146 ta 184 Oct 33 33 4.0 3.058 65.0 0-695 65.5 0.671 « 163 Aug. 36 33 41 10.8 19.3 34-2 49-4 3.040 3.003 1.890 1-547 1. 151 60.3 50.0 50.0 60.3 65.5 0.833 1. 115 1. 139 0.837 0.667 65.0 0.691 ta 185 Oct. 33 33 4.0 3.059 w 164 Aug. 36 33 4.1 10.8 19.3 34-2 3.031 1-994 1.873 1-557 65-5 50.0 50.0 65.5 0.649 1. 104 1. 105 0.656 30 49-4 1. 147 ta 186 Oct. 33 33 4.0 3.056 30 65.0 0.683 65 -5 0.654 « 165 Aug. 37 33 4.3 II. 6 19. 1 34.1 2.033 1.993 1.880 1-556 50.0 65 -5 4.0 I.IOI 0.633 3.063 give the values for the individual lines, using the quantities obtained from all of the plates belonging to the same mean latitude. These are given in Table III. In accordance with the usual notation f denotes daily angular motion, and here, of course, corresponds to the sidereal period of rotation. 36 ROTATION OF SUN TABLE III * A No. of Plates vlcm i 0?^4 4106,^^ 21 2.034 2.046 1 A ^ ■» j-y ••••.■••. 4197 •257 21 14. 14 52 4203 730 21 2.061 14 65 4207 .566 21 2.051 14 56 4216 .136 21 2.042 14 SO 4220 509 21 2.058 14 62 4232 .887 21 2.066 14 68 4233 .328 21 2.054 14 S9 4257 815 21 2.076 14 •7S 4258 •477 21 2.069 14 71 4265 .418 21 2.060 14 63 4266 081 21 2.074 14 74 4268 915 21 2.072 14- 72 4276. 836 21 2.066 14 .67 4283. 169 21 2.070 14 ■70 4284 838 21 2.069 14 69 4287 566 21 2.065 14 .66 4288 310 21 2.066 14 67 4289 525 21 2.070 14 70 4290 •377 21 2.061 14 .64 4290 •54a 21 2.070 14 70 4291 630 21 2.071 14 •71 4.08 4196 4197 .699 257 IS 15 2.023 2.026 T A -^ ■ ^*"*^ ■••«-•••■ 14 14 .40 .42 4203 730 15 2.034 14 .48 4207 .566 15 2.034 14 .48 4216 .136 15 2.028 14 •43 4220 ■509 15 2.047 14 •57 4232 .887 IS 2.042 14 •54 4233 .328 IS 2.049 14 ■58 4257 .815 IS 2.052 14 .60 4258 ■477 IS 2.044 14 ■55 4265 .418 15 2.041 14 •54 4266 .081 15 2.048 14 •57 4268 915 IS 2.042 14 ■.S3 4276 .836 15 2.039 14 ■52 4283 .169 15 2.037 14 •50 4284 .838 15 2 035 14 .48 4287 566 IS 2.032 14 .46 4288 ■ 310 15 2.033 14 •47 4289 ■525 15 2.037 14 49 4290 377 IS 2.032 14 •46 4290 •542 IS 2.040 14 •52 4291 630 IS 2.049 14 •58 II .16 4196 .699 12 Z.986 14 •37 4197 •257 12 1.987 14 •37 4203 730 12 2.005 14 .48 4207 .566 12 1-997 14 ■44 4216 .136 12 1. 991 14 •41 4220 509 12 2.006 14 49 4232 .887 12 2.001 14 •47 4233 ■ 328 12 2.003 14 .48 4257 ■815 12 2.006 14 49 4258 -477 12 2.006 14 •49 27 lO WALTER S ADAMS TABLE III— Continued ^ J^ NcofPUtes vkm ^ ii?i6 4265.418 12 2.004 I4?48 4266 081 12 2.004 ' 14.48 4268 915 12 2.001 14.47 4276. 836 12 2.010 14.55 4283 169 12 1.999 14.45 4284. 838 12 1-995 14.43 4287. 566 12 1-997 14.43 4288. 310 12 1.998 14.44 4289 ■525 12 1-999 14.45 4290 377 12 1.976 14.30 4290 ■542 12 2.001 14.43 4291 .630 12 2.000 14.44 14.86 4196 .699 18 1.928 14.16 • 4197 ■257 18 1.929 14.17 4203 ■730 18 1.946 14.29 4207 566 18 1.940 14.25 4216 .136 18 1.944 14.29 4220 509 18 1.946 14.29 4232 .887 18 I 951 14.32 4233 328 18 1.948 14.30 4257 815 18 1.957 14.36 4258 ■477 18 1-954 14.35 4265 .418 18 1.948 14.30 4266 081 18 1.949 14.31 4268 ■915 18 1-952 14.33 4276 .836 18 I -951 14.33 4283 .169 18 I -951 14.33 4284 .838 18 1.942 14.26 4287 .566 18 1.946 14.29 4288 ■ 310 18 1-954 14.35 4289 ■525 18 1.952 14.33 4290 •377 18 1.935 14.21 4290 ■542 18 1.956 14.36 4291 .630 18 1.950 14.31 19.21 4196 .699 14 1.862 14.00 4197 .257 , 14 1.862 14.00 4203 •730 14 1.874 14.09 4207 .566 14 1.878 14.12 4216 .136 14 1.865 14.02 4220 ■509 14 1.878 14.12 4232 .887 14 1.878 14.12 4233 .328 14 1. 881 14.14 4257 .815 14 1.879 14.13 4258 •477 14 1.875 14.09 4265 .418 14 1.865 14.02 4266 081 14 1.876 14.10 4268 915 14 1.872 14.06 4276 836 14 1.866 14.02 4283. 169 14 1.863 14.00 4284 838 14 1.864 14.01 4287. 566 14 1.865 14.02 4288. 310 14 1.869 14.04 4289. 525 14 1.866 14.03 4290.377 14 1.862 14.00 28 ROTATION OP SUN II TABLE Ill-Continued ^ A No. of Plates v km i I9?2I 4290.542 14 I .872 i4?o8 4291 630 14 I .875 14.09 39.66 4196 699 16 I 652 1350 4197 2S7 16 I 655 1352 4203. 730 16 I .678 13-70 4207. 566 16 I .666 13.61 4216. 136 16 I 647 13.46 4220. 509 16 I. 671 13-65 4232. 887 16 I 670 13.64 4233 328 16 I 663 13.58 4257 815 16 I 676 13.69 4258. 477 16 I 677 13.69 4265 . 418 16 I 666 13.61 4266 o8i 16 I 674 13.66 4268. 915 16 I 679 1371 4276. 836 16 I 673 13.66 4283 169 16 I 672 13-65 4284. 838 16 I 657 13-53 4287. 566 16 I 672 13 65 4288 310 16 I .676 13.69 4289. 525 16 I .676 13.69 4290 377 16 I .666 13.61 4290 542 16 I .683 13.75 4291 630 16 I .685 13.77 34.11 4196 699 IS I ■SSI 13.30 4197 257 15 I '549 13.28 4203. 730 15 I 566 13.42 4207. 566 15 I ■565 13.41 4216. 136 15 I 559 13.31 4220 509 15 I •572 13.48 4232 887 15 I 564 13.40 4233 328 15 I 565 13.41 4257 • 81S 15 I 572 13.48 4258. 477 IS I 565 13.41 4265. 418 15 I 560 13-38 4266. 081 15 I 565 13-41 4268. 915 15 I .562 13.39 4276. 836 IS I SS8 13.33 4283. 169 IS I 556 1331 4284. 838 IS I 560 13.38 4287. 566 15 I' 558 13.36 4288 310 15 I 563 13.40 4289 525 15 I 561 13.37 4290. 377 15 I .560 13.36 4290 542 15 I .568 13.41 4291 630 15 I •572 13.48 44.69 4196 699 17 I .276 12.74 4197 257 17 I .280 12.78 4203 ■730 17 I .289 12.87 4207 .566 17 I .284 12.82 4216 .136 17 I •273 12.71 4220 509 17 I .290 12.88 4232 .887 17 I .282 12.80 4233-328 17 I .288 12.86 29 12 WALTER S. ADAMS TABLE ni—CofUinued ^ J^ No. of Plates vkm ( 44^69 4257-815 17 1.300 i2?98 4258 477 17 I 295 ".93 4265 418 17 1.284 12.82 4266. 081 17 1.298 12.0 4268. 915 17 1.293 13.91 4276 836 17 1.28s 12.83 4283. 169 17 1.288 13.86 4284. 838 17 1.283 13.81 4287. 566 17 1.292 12.90 4288. 310 17 1.286 12.84 4289. 525 17 1.298 12.96 4290. 377 17 1.278 13.76 4290. 542 17 I 295 12.93 4291. 630 17 I 295 12.93 4956 4196. 699 16 1.132 ".39 4197- 257 16 1. 131 12.38 4203. 730 16 1. 145 12.54 4207 566 16 1. 146 12.54 4316. 136 16 1. 136 12.44 4220. 509 16 I 153 12.62 4232 887 16 1. 148 12.56 4233 328 16 1. 148 12.56 4257 815 IS 1. 143 12.50 4258 477 15 1. 147 "•55 4265 418 16 1. 142 12.50 4266. 081 16 1. 148 12.56 4268. 915 16 1. 146 12.53 4276 836 16 1. 145 12.52 4283 169 16 1. 145 12.53 4284 838 16 1. 146 12.54 4287 566 16 1. 143 12.51 4288 310 16 1. 149 12.57 4289 525 16 1. 152 12.61 4290 377 16 1. 140 12.48 4290 542 16 1. 152 12.60 4291 .630 16 1. 154 12.63 60.00 4196 .699 22 0.796 II. 31 4197 -257 22 0.798 "•33 4203 730 33 0.811 "•57 4207 .566 22 0.808 "54 4216 136 32 0.802 11.40 4220 509 32 0.813 11.52 4232 .887 32 0.807 "45 4233 .328 22 O.811 11.52 4257 .815 22 0.821 11.66 4258 •477 22 O.811 11.52 4265 .418 22 0.812 "•53 4266 .081 33 0.811 11.52 4268 ■915 32 0.818 11.62 4276 836 22 0.817 II. 61 4283 .169 22 0.815 11.60 4284 838 22 0.812 11.52 4287 566 22 0.815 11.56 4288.310 22 0.814 "•55 30 ROTATION OF SUN 13 TABLE III— Continued ^ A No. of Plates vkm i 6o?oo 4289.525 22 0.812 "?53 4290.377 22 0.805 11.44 4290.542 22 0.816 "•59 4291.630 22 0.817 11.60 65.04 4196.699 0.666 11.20 4197.257 0.669 11.25 4203.730 0.678 11.40 4207.566 0.674 ^^'33 4216.136 0.672 11.27 4220.509 0.684 II. 51 4233.887 0.680 11.44 4233-328 0.683 11.49 4257-815 0.683 11.49 4258.477 0.680 11.44 4265.418 0.680 11.44 4266.081 0.685 11.52 4268.915 0.689 11.59 " 4276.836 0.683 11.49 4283.169 0.680 11.44 4284.838 0.683 11.49 4287.566 0.680 11.44 4288.310 0.682 11.48 4289.525 0.688 "•57 4290.377 0.678 11.40 4290.542 0.684 11.50 4291.630 0.686 "53 74.90 4196.699 0.382 10.41 4197.257 0.386 10.53 4203 . 730 0.395 10.76 4207.566 0.389 10.60 4216.136 0.388 10.57 4220.509 0.401 10.93 4232.887 0.399 10.88 4233-328 0.388 10.57 4257.815 0.407 11.09 4258.477 0.401 10.93 4265.418 0.399 10.88 - 4266.081 0.405 11.04 4268.915 0.407 11.09 4276.836 0.403 10.99 4283.169 0.403 10.99 4284.838 0.409 II. 14 4287.566 0.401 10.93 4288.310 0.403 10.98 4289.525 0.399 10.88 4290.377 0.394 10.73 4290.542 0.405 11.04 4291.630 0.402 10.99 79.16 4196.699 0.268 10.10 4197.257 0.270 10.18 4203.730 0.271 10.20 4207.566 0.273 10.25 4216.136 0.263 9.90 4220.509 0.268 10.10 31 14 WALTER S. ADAMS TABLE m—ConUnued ^ A No. of Plates vkm f 79-16 4332.887 0.363 9?98 4233-328 0.371 10.19 4257.815 0.383 10.63 4258.477 0.369 10.12 4365.418 0.279 10.50 4366.081 0.382 10.60 4268.915 0.278 10.47 4276.838 0.377 10.42 4283.169 0.275 10.38 4284.838 0.272 10.25 4287.566 0.275 10.37 4388.910 0.282 10.63 4389.535 0.281 10.56 4390.377 0.272 10.25 4390.543 0.272 10.24 4391.630 0.279 10.49 The behavior of individual lines will be best shown if we form the difiference in the value of { between each line and the mean of all the lines. These differences are found in the short summary which follows: TABLE IV A 0?3 4?I ll?a «4?9 19? a 29^7 34? X 44?7 49?6 6o?o 65?o W®9 79?« 4196.699 -0?2 -o?i -o?i -o?i -o?i -o?i -o?i -o?i -o?i — 0?2 -0?2 — o?4 -o?a 4197 257 — O.I — O.I — O.I — O.I — O.I —O.I — O.I — O.I — 0.2 — 0.2 — 0.2 -0.3 —0.1 4203 . 730 0.0 0.0 0.0 0.0 0.0 +0.1 0.0 0.0 0.0 + 0.1 0.0 — O.I -0.1 4207.566 — O.I 0.0 0.0 — O.I +0.1 0.0 0.0 0.0 0.0 0.0 — O.I —0.2 —0.1 4216.136 — O.I — O.I 0.0 0.0 0.0 —0.2 0.0 — 0.2 — O.I — O.I — 0.3 -0.3 -0.4 4220.509 0.0 +0.1 0.0 0.0 +0.1 0.0 + 0.1 0.0 + 0.1 0.0 + 0.1 +0.1 —0.2 4232.887 0.0 0.0 0.0 0.0 +0.1 0.0 0.0 + 0.1 0.0 —O.I 0.0 0.0 -0.3 4233.328 — O.I +0.1 0.0 0.0 +0.1 — O.I 0.0 0.0 0.0 0.0 0.0 — O.I —0.1 4257-815 +0.1 +0.1 0.0 +0.1 + 0.1 +0.1 +0.1 + 0.1 0.0 + 0.1 + 0.1 +0.2 +0.3 4258.477 +0.1 0.0 + 0.1 0.0 0.0 +0.1 0.0 + 0.1 0.0 0.0 0.0 +0.1 —0.2 4265.418 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 + 0.2 4266.081 +0.1 +0.1 0.0 0.0 0.0 0.0 0.0 + 0.1 0.0 0.0 + 0.1 +0.2 + 0.3 4268.915 +0.1 0.0 0.0 0.0 0.0 +0.1 0.0 + 0.1 0.0 + 0.1 + 0.2 +0.2 + 0.2 4276.836 0.0 0.0 + 0.1 0.0 0.0 0.0 0.0 0.0 0.0 + 0.1 0.0 +0.1 + 0.1 4283.169 +0.1 0.0 0.0 0.0 — O.I 0.0 0.0 0.0 0.0 + 0.1 0.0 +0.1 + 0.1 4284.838 0.0 0.0 0.0 0.0 0.0 — O.I 0.0 0.0 0.0 0.0 0.0 +0.3 0.0 4287.566 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 + 0.1 + 0.1 4288.310 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 +0.1 + 0.3 4289.525 +0.1 0.0 0.0 0.0 0.0 +0.1 0.0 + 0.1 + 0.1 0.0 + 0.1 0.0 + 0.3 4290.377 0.0 — O.I — O.I 0.0 — O.I 0.0 0.0 — O.I — O.I — 0.1 0.0 — O.I — O.I 4290.542 0.0 0.0 0.0 + 0.1 0.0 +0.1 +0.1 +0.1 +0.1 + 0.1 + 0.1 +0.2 — O.I 4291.630 +0.1 +0.1 0.0 0.0 0.0 +0.1 +0.1 +0.1 +0.1 + 0.1 + 0.1 +0.2 +o.a 32 ROTATION OF SUN 1$ In Table IV the values at 79? 2 are based upon only seven plates, and hence are entitled to about one-half the weight of the other deter- minations. An inspection of these results shows that the lines X 4196, X 4197, A 4216, and A 4290.38 give systematically low values of f, while X 4257, A 4266, A 4291, and possibly A 4268 give high values. A simi- lar result was foimd for the first six of these lines in 1906-1907, and the general agreement of the two sets of determinations is sufficient to make the reality of these differences very probable, particularly in the case of the first three lines for which the differences are largest. Both series of observations give in the lower latitudes a value of — o?i2 for the lanthanum line A 4196, and — o?io for the mean of the two cyanogen lines A 4197 and A 4216. Since these lines originate at a low level in the solar atmosphere they furnish additional evidence for the conclusion based on the special studies of Ha and A 4227 of calcium that the rate of angular rotation increases with increase of height above the sun's photosphere. A comparison of the present results with those of 1906-1907 shows, however, that there is no such marked increase toward higher latitudes in the size of the deviations for these lines from the mean as was indicated by the earlier observations, although there is still some tendency in this direction. In the previous paper I ascribed this mainly to the effect of errors of measurement on the value of the angu- lar velocity, an effect which in the higher latitudes becomes very large. At 75°, for example, a difference of o.oi km in linear velocity corresponds to o?27 in the daily angular motion, and at 80° to o?4i. There is still sufficient evidence, however, to warrant the presumption that the retardation becomes somewhat greater in higher latitudes, and this seems to be in harmony with the conclusions derived from investigations upon Ha and A 4227 of calcium. Of the other lines which give systematic deviations, A 4257, A 4266, and A 4290.38 gave similar results in the observations of 1906-1907, although the values were as a rule somewhat larger than those foimd here. The line A 4290.38 was referred to at that time as of particular interest, on account of being an "enhanced" line of titanium, and the additional evidence confirming its behavior is especially important. The remaining two lines A 4268 and A 4291 showed no 33 l6 WALTER S. ADAMS marked tendency toward large residuals, although giving small values of the same sign as those found above. The line A 4291 is much strengthened at the limb of the sun. The cyanogen line A 4207, though showing a marked tendency toward negative residuals, gives considerably smaller values than do the other two cyanogen lines X 4197 and A 4216. This line is the most difficult of measurement of any in the list, and a part of the dis- crepancy may be due to this cause. In Rowland's general table it is ascribed to Fe^ but in the table of corrections this is changed to C. The appearance of the line seems to favor a double origin, its breadth being considerably greater than that of either A 4197 or A 4216. If it is due in part to Fe the discrepancy would be fully explained. The "enhanced" line of Fe A 4233.33 gives a value which in the mean agrees exactly with the average from all of the lines. Like the "enhanced" line of Ti A 4290.38, this line shows a considerable shift in position at the limb of the sim as compared with the center, which points to a relatively low-level origin for its "center of gravity." This result is in agreement with the rotation values. As opposed to this, however, is the occurrence of both of these lines in the chromo- sphere, the line A 4233 being one of the most prominent in this region of the spectrum. The study of "enhanced" lines under pressure is likely to throw some light upon the behavior of these lines in the sun, but at present the contradiction seems to be marked. Two other lines in the list deserve especial comment. These are A 4283 and A 4289, both due to calcium. The results indicate that slightly larger angular velocities are given by these lines, but the average difference is small, and the agreement with the result for the general reversing layer is close. We know from other considerations that the height attained in the sun's atmosphere by the gas giving rise to A 4227 is very much greater than by that which produces most of the calcium lines of moderate intensity. Accordingly we should expect a marked difference in velocity. We are obliged, however, to draw a similar conclusion in regard to the two lines in the less refrangible part of the spectrum investigated by M. Perot by inter-, ference methods.* These lines, A 5350 and A 6122, were found to show very much greater angular velocities than the general reversing X CompUs Rendus, 147, 340, 1908. 34 ROTATION OF SUN 17 layer and but slight equatorial acceleration. This result agrees much more closely with what I have found in the study of A 4227, a discus- sion of which will be given at a later point in this paper. The line A 6122 is one of the strong lines belonging to the red triplet in the second sub-series of calcium. The other lines do not belong to known series^ although both X 4283 and X 4289 have certain definite relations with other lines in their vicinity. The three lines of shorter wave- length do not seem to be present in the chromosphere, at least in any considerable intensity, but A 6122 is given by Young in the revision of his chromospheric list.' We may now pass on to a consideration of the general results. If mean values are formed from Table II and grouped about thirteen points of latitude we obtain a summary of the following form. The weights are given according to the number of observations. TABLE V ♦ Weight V km ^ Period, Days o?3 21 2.063 14^65 24-57 4.1 15 2.040 14.52 24.79 II. 3 12 1.992 14.44 24.98 14.9 18 1.944 14.28 25.21 19.2 14 1.868 14.04 25.64 29.7 16 1. 671 13.66 26.35 34.1 15 1-557 13-39 26.97 44-7 17 1.283 12.81 28.10 49.6 16 1. 141 12.49 28.82 60.0 22 0.811 11.52 31 25 65.0 17 0.678 II. 41 31-55 74.9 17 0.398 10.84 33.21 79.2 7 0.274 10.29 35 26 The velocities given in this table have been plotted graphically, and the full line shown in Fig. i indicates the curve to which they correspond most closely. The general agreement of the points is good, the largest deviation from the curve falling at the point of lowest weight 79? 2. Near 50° there is a distinct point of inflection. The curve shown by the broken line in the same figure represents the observations of 1906-^1907. For numerical comparison the following table, giving the values from the two series of observations for every 5^ of latitude, will be found useful. They are taken from large-scale > Schdner's Astronomical Spectroscopy (Frost), p. 422. 35 i8 WALTER S. ADAMS * a/ A 1 « ti M M II Jl i? 4 // * / 1 /. / * / f / / X / V / / J / / * If il I o % S> % % 8 ■ o v8 O o ■ o 00 r* H Iff o o O ^ •0 c d X c C 6 8 •0 M Ik •o M o ■ o 36 ROTATION OF SUN 19 drawings of the two curves, and are quite accurate enough for the purpose. TABLE VI ^ V (1908) V (i9o6-7> n^9os) f (igo6-7) km km o«> 3.060 2.077 14^63 14^75 5 2.043 2.056 14 56 14.65 10 2.000 2.012 14.42 14.50 15 1-943 1-954 14.28 14.36 ao 1.863 1.872 14.08 14.14 25 1. 761 1.768 13-80 13-85 30 1.659 1.660 13.60 13.61 35 1-537 1-532 13-32 13.28 40 1. 414 1.406 13.10 13-03 45 1.274 1.274 12.79 12.79 50 1. 126 I -137 12.44 12.56 55 0.967 0.991 11.97 12.27 60 0.820 0.851 11.64 12.08 65 0.677 0.711 "-37 11.94 70 0-533 0.569 11.06 II. 81 75 0.396 0.429 10.86 11.77 80 0.266 0.290 10.88 11.85 A comparison of the two sets of observations shows slightly larger values for the earlier series between latitudes 0° and 25®, practical coincidence between 25° and 50®, and decidedly larger values again for the earUer series above this point. The differences in the lower latitudes reach a maximum of 0.017 km at the equator, and have an average value of about o.oi km throughout the first 25° of latitude. These quantities, though larger than would be expected from errors of measurement, are still so small as to give little indication of the presence of systematic errors, or to show a possible variation in the sun's rate of rotation. The matter is different in the higher latitudes, however. At 50® the two series of observations begin to separate from one another, and at 70° a difference of 0.036 km is reached. Though determinations of the probable error have shown that the value is usually somewhat larger in higher latitudes where the measiured displacements are small, differences of this amount are evidently systematic, and the cause is to be sought either in a real change of rotation period in the higher latitudes, or in some soiurce of systematic error in the observations, whether instrumental, or in the sun itself. At present there seems to be no definite criterion for settling the question, but there is, I 37 20 WALTER S. ADAMS think, sufficient evidence to furnish a presumption against the reality of a change in the sun's rotation period. The most important part of this is the difficulty of conceiving a change of rotation rate which is confined to high latitudes. It would certainly seem probable that any such change of rate would be in some way connected with the sun- spot activity, and would be conspicuous in the zones of greatest spot frequency between io° and 30^ of latitude. This is the conclusion which Halm has drawn from his two series of observations,' which show large dififerences in these zones as well as in higher latitudes. The results given here indicate almost no difference between the values of 1908 and those of 1906-1907 for these zones. As against this it may be argued that evidence is gradually accu- mulating to show that the rotation rate at the sun's equator is more nearly constant than at any other latitude, and that the so-called ''equatorial acceleration" is more truly a polar retardation. In this direction, for example, point the observations of Halm showing the greatest variations in angular velocity in higher latitudes, as well as the results given here, on the one hand for the low-level lines of lan- thanum and cyanogen, and on the other hand for the high-level line A 4227 of calcium and the a line of hydrogen. All of these show the greatest differences in angular velocity toward the pole. The force of this argument is considerable, and it may perhaps account for the apparent anomaly of a change in rotation rate which is confined to high latitudes on the sun. A second argument which is entitled to considerable weight is the satisfactory agreement of the results of the 1908 observations with Faye's formula for the rotation period of sun-spots, and the fact that Dun^r's observations, as well as those of Halm, are also in excellent accord with the same formula. If we put v— (a-l-6 cos' 4) cos ^ , f=a'H-y cos» 4> , and deduce the values of the constants by solving the normal equa- tions given by the observations, we fijid ^=(1.506-1-0. 549 cos' 4) cos ^ , i=io?62-|-3?99 cos* ^ . > TrafuacHons of the Royal Society of Edinburgh, 41, Part I. 38 ROTATION OF SUN 21 In obtaining these results the observations have been assigned their respective weights, but the observation equations have been formed by multiplying with these weights instead of by their square roots. The residuals given by these equations are as follows: TABLE VII ♦ V km i o?3 + 0.009 +o?04 4-1 — 0.006 — 0.07 II. 3 — 0.003 — 0.02 14. 9 — 0.007 — 0.07 19.2 — 0.016 —0.14 29.7 + 0.003 + 0.03 34.1 0.000 + 0.03 44-7 + 0.016 + 0.17 49-6 + 0.015 + 0.19 60.0 — O.OIO — O.IO 65.0 +0.001 +0.08 74.9 — 0.004 -0.05 79.2 -0.013 -0.47 The general agreement of these results is quite as satisfactory as could be expected from a formula of such simple form, the largest residual being the point of low weight and high latitude at 79? 2. In the 1906-1907 observations, on the other hand, it was found that Faye's equation gave deviations which, though small, were system- atically negative in mean latitudes, and positive in high'latitudes, and an equation containing three constants was employed in order to obtain suitable agreement. The fact that the results of Dun^r and Halm as well as my own are satisfactorily represented by Faye's formula, while Sp5rer's formula gives systematic deviations, points to the conclusion that this simple expression represents with suflBcient accuracy the motion of the sim's reversing layer to within about 10^ of the pole, beyond which observations are lacking at present. The investigation of the zone between 80° and 90° of latitude, undertaken when the position of the sim's pole is most favorable for observations necessarily so difficult, will be of great interest, and should increase materially oiu: knowledge of the law of rotation. The close absolute agreement of Dundr's results with those given by the 1908 observations is also opposed to the idea of a change in the sun's rate of rotation. A comparison for the six latitudes employed by Dun6r gives the following summary: 39 22 WALTER S. ADAMS TABLE VIII ^ Dun^ Adams km km o?4 2.07 2.06 15.0 1.97 1-94 30.0 1 .69 1.67 45 1.27 1.27 60.0 0.80 0.81 74.9 0.40 0.40 Dun^r's results are based upon observations extending through six years, from 1887 to 1889, ^^'^ irom 1899 to 1901. Hahn in his discussion' advocating a variation in the rotation rate of the sun has analyzed these observations, using as a basis Faye's equation in the form TF=(a— 6 sin* ^) cos ^ . In this form the 1908 observations would give v=*(2. 054— 0.549 sin* ^) cos ^ . Halm's comparison of the Upsala and the Edinburgh observations on the basis of a three-year period would require a minimum value of the quantity a for the epoch 1908.5, which is about the mean date of these observations. The value foimd here, however, is prac- tically equal to the largest value obtained by Hahn in any one of the years 1901 to 1906, which included two maximum values of a. The value of b is also opposed to the idea of a progressive change in this quantity unless we assume that after decreasing from 1901 to 1905, it has reversed its direction and returned in 1908 to a value almost identical with that of 1901. The general conclusion from these considerations seems to be that it is probable that the observations of 1906-1907 were affected by slight systematic errors which amounted to as much as 0.03 km in the higher latitudes. Attention has already been called to the fact that the astigmatism of the sun's image may have influenced the values to some extent, and it is possible that chsCnges of focus may have also introduced slight errors. Another possible source of error due to disturbances on the sun itself will be discussed in a succeeding paragraph. In any event, it seems probable that the 1908 series of observations, being comparatively free from such defects of the solar > Astronomische Nachrichten, 173, 294, 1907. 40 ROTATION OF SUN 23 imagey and containing as they do a valuable check upon instrumental conditions in the exposures upon the pole of the sun, deserve a con- siderably higher degree of confidence. Determinations of the probable errors for tne observations of 1908 indicate that these are smaller than for the earlier series. From a number of plates taken at random we find at 45° of latitude: for a single line, €=±0.009 ^™ 9 or for the mean value from the plate, Co ==±0.002 km . The 1906-1907 observations gave €=±0.015 km , and Co «= ±0.004 km . In these determinations the lines giving the principal systematically large or small values have been omitted. The 1908 probable errors, accordingly, are based upon sixteen lines, and the 1906-1907 errors upon fourteen lines. The probable errors are slightly larger in the higher latitudes. THE MOTION OF THE REVERSING LAYER IN THE VICINITY OF A SOLAR VORTEX On September 15 four plates of the region of the spectrum in- cluding A 4227 were made with the rotation apparatus at latitudes ranging from 0° to 75^ in steps of 15® each. On this date two spots of considerable size, one at 6° south latitude, the other at 1 1° north latitude, were close to the west limb of the sun, the northern spot being but a few hours distant from the visible edge, while the other had just passed beyond it. Plates taken with the spectroheliograph had shown these spots to be sxirrounded by vortices in which the motion was apparently in opposite directions in the two cases. Observations made by Hale during the passage of the spots across the sun's disk also showed opposite directions of polarization for the components of the double lines in the spot spectra. The region between the two spots, as indicated by photographs taken with the Ha line, was in an extremely chaotic state, owing probably to the intermingling of the vortices. One of the settings in the case of the rotation plates fell 41 24 WALTER 5. ADAMS at latitude 14? 9 north, or 4° north of one of the spots; the other at 0°, or 6^ north of the second spot, in this case being considerably east as well. When the measurement of the plates was begun it was at once seen that these two latitudes gave extremely discordant values as compared with the normal. Accordingly, the plates were in- vestigated separately and measmres made not only on A 4227, but also on a number of the general reversing layer lines. As in the case of all of the photographs intended for the study of A 4227, the density was made great in order to facilitate settings upon this very broad line, and for this reason the list of lines regularly used for the re- versing layer could not be employed. So a number of stronger lines which appeared suitable for measmrement were selected. The individual results for these lines and for A 4227 are given in the following table: TABLE IX Number of Date Number of ^ A 4^37 Plate 1908 Lines V km vkm «I73 Sept. 15 la o?o 1. 918 1-974 14.9 1.896 1.960 29.8 1.662 1-733 44.7 1.249 1-325 62.8 0.722 0.813 75-3 0.387 0.492 W174 Sept. 15 12 0.0 1.900 I 963 14.9 1.882 1.959 29.8 1-673 1-723 44.7 i.a34 1-325 62.8 0.731 0.808 75-3 0.405 0.460 «i75 Sept. 15 12 0.0 1.954 2.062 14.9 1-913 1.984 29.8 1.664 1.722 44-7 1-257 1.282 62.8 0.717 0.815 75-3 0.377 0.460 « 176 Sept. 15 12 0.0 1.949 2.029 14.9 1.906 1.982 29.8 1.667 1.722 44.7 1.222 I 303 62.8 0.726 0.859 75-3 0.287 0.493 If we combine the preceding results we obtain the following summary. The normal value for the reversing layer is given in the third column for comparison. 42 ROTATION OF SUN 25 TABLE X * vkm Normal vkm X ^227 V km o?o 14.9 29.8 44.7 62.8 75-3 1-930 1.899 1.666 1.240 0.724 0.389 2.060 1-945 1.662 1.277 0.743 0.393 2.007 1. 971 1-725 I 309 0.844 0.476 It is seen that in latitudes 30° and more the values derived from these plates agree well with the normal values. At 15*^, however, the results are decidedly lower both for the reversing layer and for A 4227. The same is true of latitude 0°, and the differences here are even greater. The plates taken with the spectroheliograph indicate a direction of motion in the vortex surrounding the northern spot that is counter-clockwise as seen from above. Accordingly, for a position north of the spot the component of motion in the line of sight at the sun's west limb would be toward the observer, and its efifect would be to reduce the value of the linear velocity observed on the rotation plates. This is what is found. In the case of the second spot the vortical motion should be in the opposite direction, judging from polarization and spectroheliograph results, and this is opposed to what is foimd from the rotation results at latitude 0°, if the latter are affected by this vortex. An examination of the Ha photographs shows, however, that the point from which the light is taken into the rotation apparatus falls at a considerable distance from the spot, and over a region of great irregularity where the two vortices seem to mix with each other, and where no well- defined direction of motion seems to exist. In any case the observa- tions should be much more extensive and should include more points of latitude in order to give any definitive evidence as to the question of direction of rotation in such regions. The important fact is that solar vortices of this nature seem to have a marked influence upon the reversing layer, and that rotation results obtained when such regions are at the limb of the sun may be very seriously affected by the proper motion of the reversing layer. Since these vortices usually accompany spots, a natural inference would be that the greatest variations in values of the rotational velocity would be found in the latitudes where 43 26 WALTER S. ADAMS spots are most frequent, that is, in the zone lo^ to 30°. It is possible that this may to some extent influence the results found by Hahn m the two series 1901-1902 and 1903. The fact that vortices frequently exist where there are no spots, and extend to high latitudes, would also account for variations of rotational velocity found beyond the zones of principal spot activity. This may be a partial cause of the diflFer- ences in latitudes 50° to 80° between the results 1906-1907 and 1908, since neither series contains a sufficiently large number of observa- tions to eliminate the influence of a few cases of such pro[>er motion. The necessary conclusion is that in making spectroscopic observations of the rotation of the sun especial care should be taken to avoid regions of the surface which show evidences of vortical disturbance. RESULTS FOR A 4227 The study of at least one of the stronger lines of calcium was made most desirable by the results obtained for the a line of hydrogen, as well as the differences found among the various lines used in the investigation of the reversing layer. Unfortunately the lines of greatest interest, namely H and K, are practically excluded by the inunense variation in their physical appearance at different parts of the sun's disk, according to the presence or absence of caldum flocculi. They also appear to be especially subject to disturbances arising from motion of the calcium vapor in the Une of sight." Ex- cluding these lines, it was evident that the so-called "blue line" of calcium at X 4226.91 would prove most valuable for investigation. It is known to rise to a great height in the chromosphere, although, of coiurse, much lower than either H or K, and its appearance in the spectrum is such as to admit of measures of considerable accuracy. There is also a decided practical advantage in having the line fall within the region observed for the motion of the reversing layer. It at first seemed probable that the same plates which were used for the study of the reversing layer could be employed for A 4227 as well. A few measures, however, showed that the line is of such great intensity, and the wings so broad under high dispersion that especially dense negatives would have to be employed, and it is upon > CofUributions from the Solar Observatory, No. 6; Astrophysical Journal^ 33» 45-53. 1906. 44 ROTATION OF SUN 27 a series of plates taken for this line alone that the results given here are based. An interesting by-product of the study of the line is the discovery that it is double. Several plates show this fact clearly, although the actual separation is extremely small. The results for the individual plates are given in the table below. In each case the value is the mean of two measures, one by Miss Lasby, and the other by myself. TABLE XI Plate ♦ vkm ♦ vkm ♦ vkm « vkm f tf km ♦ vkm w 149 o?3 2.10 I4?6 2.01 29?6 1-75 44?9 1.32 60? I 0.84 74?9 0.47 «# 150 0-3 2.09 14.6 1.99 29.6 1.72 44.9 I 32 60.1 0.83 74.9 0.46 « 152 0.1 2.14 14.6 1.99 29.7 1.72 44.6 1.32 59.8 0.89 74.9 0.47 « 153 0.1 2.12 14.6 1.99 29.7 1-74 44.6 1-33 59-8 0.85 74-9 0.43 « 154 0.1 2.10 14.6 2.02 29.7 1.74 44.6 1.34 59« 0.81 74.9 0.43 «i57 03 2. II 14.6 2.04 295 1. 71 44.1 1.36 5«.4 0.88 72.0 0.50 w 158 03 2.10 14.6 2.04 29-5 1.70 44-1 1-34 58-4 0.88 72.0 0.48 « 167 0.4 215 14.5 2.05 29.4 1.78 44.3 1.40 59-5 0.92 74.4 0.52 w 168 0.4 2.12 14.5 2.05 29.4 1.74 44.3 1-39 59-5 0.92 74.4 0.57 » 169 0.4 215 14-5 2.06 29.4 1.74 44-3 1-39 59-5 0.89 74.4 0.57 « 170 0.4 2.16 14-5 2.05 29.4 1-75 44.3 1.42 59-5 0.97 74-4 0.54 w 188 0.1 2.10 14.8 2.09 29.7 1.78 44.9 1.36 60.2 0.90 75-9 0.48 w 189 0.1 2.12 14.8 2.02 29.7 1.78 44.9 1-39 60.2 0.94 75-9 0.48 These values give the following summary: TABLE XII ♦ A 4237 Reveksino Layer vkm 1 Period vkm 1 Period 0?2 2.12 i5?o 23 -9 2.06 I4?7 24.6 14.6 29.6 44.6 59-6 2.03 1-74 1.36 0.89 14.9 14.2 13.6 "•5 23.2 25. 3 26.5 28.8 1.95 1.67 1.28 0.82 14.3 137 12.8 "•5 25.2 26.4 28.1 31.2 74.6 0.49 131 27.4 0.40 10.8 33-2 A comparison of the results for A 4227 with those for the reversing layer shows: first, that the absolute velocity values are larger for A 4227; and second, that the decrease of angular velocity is much less marked toward higher latitudes. At the equator the difference in angular velocity is o?4, while at 45° of latitude it is o?8. The sudden increase in the value of f at 75° will be discussed more fully in connection with the results for Ha, If we assume, however, that the increase is due to accidental errors, and take values from the curve 45 28 WALTER 5. ADAMS at 60*^ and 75°, we find at 60° a difference in the value of f from that for the reversing layer of i?4, and at 75° of i?5. As will be seen later, similar results are found for Ha, Reference has already been made to the results obtained by M. Perot for two of the less refrangible lines of calcium by the aid of inter- ference methods. Preliminary values of the angular velocity given by him are as follows: TABLE XIII A ♦ -o» ♦ -45?7 5349-6 6122.4 15^1 14.7 I4?2 14.2 At the equator these results agree closely with those for A 4227. At 45°, however, they give a considerably larger value, indicating less equatorial acceleration than in the case of A 4227. This difference may be much modified by more complete results, but if it still remains it wiU indicate a higher effective level for these lines at the limb than for the more refrangible line. At first sight, in view of the great strength of A 4227 in the chromosphere, this would seem improbable, but the width and the intensity of its wings and the comparative narrowness of the central line indicate that a large part of the line is produced in a region of relatively dense calcium vapor, and so at a moderately low level. This level must, nevertheless, be above that in which the wings of the stronger iron lines in the violet are formed, since plates of the center and limb spectrum show but a slight effect for A 4227 when the wings of the iron lines are almost completely obliterated. This is in agreement with the rotation values. RESULTS FOR THE a LINE OF HYDROGEN The special investigation of the Ha line was begun, as stated in a previous paper,* in consequence of the remarkable behavior of this line at the limb of the sun. Its marked increase in width, and the absence of any displacement such as is found for the great majority of lines between limb and center, pointed to a very high-level origin, a fact, of course, well known from observations of the intensity of the > Contributions from the Mount Wilson Solar Observatory, No. 24; Astrophysical Journal, 27, 213-218, 1908. 46 ROTATION OF SUN 29 line in the upper chromosphere. Measures of the center and limb plates also indicated a considerably higher rotational velocity, and this result was confirmed by the earliest rotation plates. The preliminary values for a few of these plates were published in the paper already cited. At the same time at which these determinations were being made, measures were obtained of the motion of rotation given by the hydrogen flocculi upon plates taken with the spectroheliograph in the S line of hydrogen.* These showed a marked tendency toward a uniform rate of rotation, but the absolute rate was decidedly smaller than that derived from the spectroscopic measiu^es. Two possible explanations of this behavior at once presented themselves. The first was that Hh might give a different rate from Ha^ a conclusion by no means improbable, in view of the different way in which they are affected at the limb of the sun. The second was that the marked increase of intensity of Ha near the limb might indicate quite a differ- ent level for the spectroscopic determinations from that of the hydrogen flocculi. Photographs of the Ha flocculi were essential to test either hjrpothesis, and experiments were begun by Mr. Hale and myself with the 30-foot spectroheliograph. These when continued by Hale and EUerman with the 5-foot spectroheliograph led to the discovery of the solar vortices.* The earliest photographs of the Ha line were taken on rapid plates sensitized to the red with pinacyanol. After the publication of Wallace's sensitizing process for "Pan-iso" plates^ this was em- ployed to great advantage. It was found, moreover, that the ex- posure times on rapid plates sensitized in this way were very moderate. Accordingly it occmred to me to endeavor to obtain the advantage of finer sflver grain by the use of slower plates, and for this purpose the Seed " Process" plates were tried. These have proved very successful, giving fine grain and excellent contrast, while the ratio of sensitive- ness in the red to that of similarly treated rapid plates seems to be greater than the corresponding ratio of the unbathed plates in the blue. The advantages in the use of these plates for such a line as > CotUributions from the Mount Wilson Solar Observatory, No. 25; Astrophysical Journal, 2*j, 219-229, 1908. > Hale, Contributions from the Mount Wilson Solar Observatory, No. 26; Astro- physical Journal, 28, 100, 1908. 3 Astrophysical Journal, 26, 299-325, 1907. 47 30 WALTER 5. ADAMS Ha are considerable, since the edges of the line are at best poorly defined, and the finer grain and the superior contrast assist materially in the measurement of the line. It is probable that the later plates show a considerably higher degree of accuracy for this reason, and in the summaries of the results which follow, each series begins with the first photographs taken on the slower plates. It was foimd early in the study of the Ha line that its width changes very rapidly within a small distance of the sim's limb. In fact, the entire variation in appearance between limb and center seems to take place within a distance of little more than one-thirtieth of the solar radius. It seemed probable that this effect might be due to the relative level in the two cases, the effective level of the line at the limb being higher than that inside the limb. If this is the case, we should expect the effect to show in the rotation values given by the line in the two positions. Accordingly two series of observations have been made on Ha\ the first at points close to the limb; the second at points averaging about 3 nmi inside the limb. The values obtained from the latter have, of course, been reduced to the limb by correcting for this distance. The results for the individual plates of Ha near the limb follow. As in the case of A 4227 each result is based on two measures. TABLE XIV Plate ♦ vkm « vkm ^ vkm ^ vkm ^ vkm ^ vkm » no 0?2 2.12 iS?i 2.04 29?9 1.7s 44^3 1-45 6o?4 0.97 75?2 0.53 « 115 0.4 2.13 14 .6 2.03 29.6 1.72 44 •9 1.38 60 •5 0.94 759 0.49 w ii8x 0.6 2. IS 14 .4 2.05 29.4 1-77 44 ■9 1.43 60 .4 0.98 75.9 0.49 » Il8a 0.6 2.17 14 .4 2.06 29.4 1-74 44 9 1.44 60 .4 0.99 75-9 0.52 » 122 0.5 2.13 14 •S 1.99 29s 1.79 44 •S I -as S9 ■S 0.93 75-1 0.49 w 123 o-S 2.16 14 •s 2.03 295 1.78 44 •S 1.41 S9 ■5 0.96 75-1 0.48 « 124, o-S 2.10 14 ■s 2.09 29s 1.78 44 •s 1. 41 59 •5 0.99 75-1 0.53 « 1 24a 0.5 2.14 14 •s 2.05 295 1.78 44 5 1.39 S9 ■5 0.97 75-1 0.52 « 125, 0.5 2.17 14 s 2.05 29s 1.82 44 S 1.42 S9 5 0.97 75 0.50 « 125, 0.5 2.14 14 ■s 2.02 295 1.80 44 5 1-43 S9 •5 0.96 75 0.49 w 1 261 0.5 215 14 ■s 2.04 29s 1. 81 44 5 1. 41 59 5 0.9s 750 0.50 w 126a 05 2.16 14 s 2.02 295 1.84 44 S I 42 59 5 0.98 75 -o 0.53 w I27x 0.5 2.16 14 s 2.04 29s 1.80 44 S 1.38 59 ■5 0.95 74.5 0.52 « 127, 0.5 2.17 14 s 2.02 29s 1.83 44. 5 1.40 59 5 0.98 74.5 0.50 w 129 o-S 2. IS 14 s 2.04 29s 1.82 44. S 1.41 59 5 0.96 74.5 0.52 « I30X o-S 2-IS 14 5 2.03 29s 1.80 44 S 1.40 59 5 1. 00 74.5 0.53 « i3o« 0.5 2.14 14. 5 2.03 29s 1.79 44 S 1.40 59 5 0.95 74.5 0.5s « 131 OS 2.14 14. S 2.04 29s 1. 81 44 5 1. 41 59- 5 0.98 74-5 0.52 «# 141 0.1 2.14 14. 8 2.04 29.7 1.79 44. 9 1. 41 60. I 0.99 75-1 0.57 «I44 0.1 a. IS 14. 8 2.04 29.7 1.79 44 9 1.44 60. I 1. 00 751 0.59 w 171 0.3 2.12 IS-2 2.04 30.0 1.76 4SO 1.4a 60.2 0.98 75 -o 0.60 48 ROTATION OF SUN 31 If we form means from the above results we obtain the following summary. The corresponding results for the reversing layer are given in the last three columns. TABLE XV * Ha RxvEKsmc Layer vkm 1 Period vkm ^ Period o?4 14.6 29.6 44.6 59.8 75 -o 2.15 2.05 1.79 1. 41 0.97 0.52 15^2 150 14.6 14.0 13-7 14 3 23.6 24.1 24.7 25 -7 26.2 25.2 2.06 1-95 1.67 1.28 0.81 0.40 I4?6 14.3 13.7 12.8 "5 10.8 24.6 25.2 26.3 28.1 31.2 33.2 Two important conclusions are evident from an inspection of these results. The first is that the hydrogen gas giving rise to Ha moves at a much higher velocity than the reversing layer. The second is that the law of change of velocity with latitude is very different, the equa- torial acceleration being comparatively slight. Both of these con- clusions were derived from the preliminary values given in the paper previously referred to, but the absolute values differ considerably. In the preliminary results, for example, no equatorial acceleration whatever was found, while these results give a difference of over 1° in the value of f between the equator and high latitudes. The absolute velocity at the equator was also foimd to be about o?3 larger in the preliminary results. The greater part of this difference is no doubt due to the small number of plates upon which the early measures were based, and also to the fact that these plates were comparatively inferior in quality, being obtained previous to the use of the finer- grained "Process" plates. There is, however, another cause which may contribute to a systematic difference in the two cases. This is the fact that the earlier plates were taken at points closer to the edge of the sun than were the later plates, so close, indeed, that on two out of the three Ha plates used in the early measures, the bright chromo- spheric line appears in addition to the dark line. Accordingly it is altogether probable, judging from results found for Ha inside the limb, that in the plates used in the preliminary determination a higher average level was in question than in the case of the plates 49 32 WALTER S. ADAMS discussed here, and that this higher level would give larger velocities and less equatorial acceleration. The results obtained from plates of Ha taken at points within the limb are given below. TABLE XVI Plate « vkm ^ vkm ^ V km * tf km ♦ vkm ♦ vkm w ii6 o?i 2.14 14^9 2.03 29-9 1.72 4S?i 1.38 6o?o 0.91 76?2 0.44 h 5* ROTATION OP SUN 35 agreeing closely with those of 1906-1907 between latitudes 0° and 50®. Above 50® they give larger values, the difference in linear velocity reaching at a maximum 0.036 km. 2. The general agreement of the results, and the excellent accord with Dim^r's values, are opposed to the existence of a variation in the rotation rate between 1906 and 1908. If any such variation exists it is confined to the higher latitudes, and does not appear in the zones of greatest spot activity. The results are also opposed to a three-year period of variation, such as was obtained by Halm from a comparison of his values with those of Dim^r. 3. The observations of 1908 confirm those of 1906-1907 in show- ing that different lines give different velocities. Lines of lanthanum and cyanogen give low velocities; certain lines of manganese and iron give high velocities. The investigation of two "enhanced" lines indicates a tendency toward low values for lines of this t)rpe. In one of the cases this effect is very marked. Lines considerably strength- ened at the Sim's limb give high values in general. 4. When lines give systematically large or small values for the rotational velocity the differences from the mean become greater toward higher latitudes. 5. The results given by the 1908 observations are satisfied within the limits of accidental error by the equation given by Faye for the motion of the sun-spots observed by Carrington. The fact that the observations of Dun^r, Halm, and myself are all satisfied by this equation indicates that this represents the law of rotation of the sim's reversing layer to within at least 10® of the pole. 6. Comparison of the probable errors for the two series of obser- vations indicates a substantial gain in acciu'acy of measurement for the 1908 series over that of 1906-1907. 7. The motion of the reversing layer in the vicinity of solar vortices may be seriously influenced by the motion of the vortices, and the rotation velocities obtained from such regions are subject to large systematic errors. It is important in taking observations for rotation to avoid all such disturbed areas. 8. A special study of the calciiun line A 4227 shows that the rotational velocity derived from this line is higher than that for the general reversing layer, the difference at the equator amounting to o?3. Also the decrease of velocity with increasing latitude is much 53 36 WALTER S. ADAMS less marked than for the reversing layer. At 75^ of latitude the angular velocity of X 4227 is i?5 greater than for the reversing layer. 9. A special study of the a line of hydrogen shows that the rota- tional velocities which it gives depend upon the distance from the Sim's limb. Results obtained from Ha at the limb of the sun are considerably larger than those for the reversing layer and show a comparatively small decrease in the value of the angular velocity toward the pole. At the equator the difference from the reversing layer amoimts to o?6, and at 75^ of latitude to 3?o. 10. At a distance averaging 35'' inside the sun's limb the results obtained from Ha are considerably smaller than the corresponding values at the limb, although still much larger than for the reversing layer. They also average somewhat larger than for A 4227. 11. The sudden increase in angular velocity at latitude 75° may perhaps be a genuine effect similar to that foimd among the lines of the reversing layer which give systematic deviations. 12. The large rotational values given by A 4227 and Ha, and the differences found for Ha at the limb and inside the limb, may be ac- counted for on the basis of differences of level in the solar atmosphere. It may be of interest, acting on a suggestion from Professor Kap- teyn, to call attention to the fact that the difference in the values of the rotation obtained from different spots on the planet Jupiter may perhaps be explained on a basis similar to that given for different elements in the sun. A difference of about 6 minutes has been found in the rotation period given by different spots upon the surface of Jupiter.^ The ratio of this difference to the longer rotation period is of much the same order as that obtained by comparing A 4227 with the reversing layer, or Ha with A 4227. It seems reasonable to conclude, therefore, that the different velocities foimd in JupUer^s atmosphere may be wholly accounted for upon the basis of a dif- ference of level for the various spots observed. I wish to express my appreciation to Miss Lasby for her careful and accurate measiu'ement of the majority of the plates included in this discussion. MouKT Wilson Solar Observatory November 1908 > Stanley Williams, Monthly Notices^ 56, 143-15 1, 1896. 54 » 1 r I / ~..0, ^ r-tO.»36'0 ■ cjontributions from the mount wilson solar Observatory NO. 34 A ON THE SEPARATION IN THE MAGNETIC FIELD OF SOME ONES OCCURRING AS DOUBLETS AND TRIPLETS IN SUN-SPOT SPECTRA . BT ARTHUR S. KING I • Reprinted from the Astropkyntal Journal^ Vol. XXIX, Janjuary 1909 .jA'X. * * I • '^rur ^ Contiibutioiis from the Mount Wilson Solar Observatory, No. 34 Reprinted from the AstrophysiciU JcprruU, Vol. XXIX, pp. 76^83. 1909 ON THE SEPARATION IN THE MAGNETIC FIELD OF SOME LINES OCCURRING AS DOUBLETS AND TRIPLETS IN SUN-SPOT SPECTRA By ARTHUR S. KING In the "Addendum" to Professor Hale's paper " On the Probable Existence of a Magnetic Field in Sun-Spots,"' it was stated that a number of the iron and titanium lines appearing as doublets in the spectra of sun-spots appeared as doublets also in the spectnun of the spark in the magnetic field when observed at right angles to the lines of force, the sim-spot triplets also retaining their character when observed in this manner. This of course referred to the appearance of the lines under the dispersion then used, leaving 0|>en the question whether they would show a higher order of sepa- ration when studied with a view to determining this. The next step was to observe the polarization of the components, and this has brought out the fact that all of the doublets listed in the ' 'Addendum" are in reality quadruplets, and the list has been much extended, especially on the side of the titanium spectrum, which has been photo- graphed through the range from A 4400 to A 6400, not only with the light parallel to the lines of force and also at right angles to them, but in addition a set of photographs has been obtained in which the light at right angles to the lines of force was passed through a Nicol prism in two positions 90° apart. This operation separated the light vibrating parallel to the force lines, producing the middle component of the normal triplet, from the light whose vibrations are in a plane perpendicular to the lines of force, giving the side com- ponents of the normal triplet. The great majority of the iron and titanium lines show the regular behavior of the normal triplet, the middle component remaining single; but for the lines appearing as doublets in sim-spots, the Nicol when tiuned so as to transmit > Contributions from the Mount Wilson Solar Observatory , No. 30; Astrophysical Journal, a8, 315-343. iQoS* 55I I ARTHUR S. KING the light with vibrations parallel to the lines of magnetic force gives the middle component, in all but two cases, either as a doubled line or one of such widened and diffuse appearance that evidently only a lack of sufficiently high dispersion stands in the way of a dear resolution of the components. The spot triplets retain in all cases their sharp middle component when examined in the laboratory at right angles to the lines of force. The following table gives the character of each line in the spark with the measurements of its components. The last column, giving the appearance of the Une in sun-spots and measurements when these could be made, was kindly contributed by Mr. Adams. Most of the measurements of spark lines were made by Miss Wickham. The magnetic field usually employed with the titanium spark was 12,500 gausses; for iron about 15,000 gausses. When a stronger field was employed for the separation of difficult lines, as in the blue of titanium, the measurements were reduced to the standard field. As the separation of the lines given at right angles to the field by the light vibrating in a plane perpendicular to the lines of force is the same as that given by the light observed parallel to the lines of force, photographs taken in the latter way were sometimes used when better measurements could be thus obtained; though for most of the titanium spectrum good photographs were available for the two positions of the NicoL IRON X SpttkObienred •t Right Ai^et to Force LuieB 6173.55 Triple 6213.14 6256.57 Quadruple Quadruple 630X.72 Quadruple 6302.71 Triple 6337.05 Quadruple AA Spark, Trmna- venc Effect* Not separated 0.460 0.438 0.287 (scarcely resolved) Not separated 0.431 AASpuk, Longitodiiud Effect* Chancter In Sna-Spoti 1.094 0.840 Narrow. Com- ponents diffuse 0.792 1. 144 0.880 Wide triplet Central line about I intensity of side components Wide doublet AX >« 0.136 Widened 3.t No evidence of doubling Wide doublet AX-0.138 Wide triplet Central line about i intensity of side com- ponents Wide doublet AX * o . 1 72 *T1ieee coIiiBiat refer to tbe eperk in the magnetic field wfaca obsenred acroae the linei of force and aloag them re^iecd^y. tThe eetimatei of videninf axe on an arbitrary scale of o to 5. s« SEPARATION OF SUN-SPOT DOUBLETS TITANIUM A SpaikObserved at Rigfat Ancles Co Force Luics AA Spark, Trans- TRve Effect* AASpaik, Loagitndinal Effect* Character in Sun-SpoCs 4440.49 4444-73 4450.65 Quadruple Quadruple Sextuple 0.144 0.189 0.189 0.182 0.202 0.245 (mean for two pairs) Narrow. Strengthened but scarcely widened Narrower than in disk and probably slightly weakened Possibly dightfy strengthened; not widened. Enhanced line 4464.62 Quintuple 0.213 ( 0.204 (0.214 0.460 Slightly weakened and nar- rowed 4471-02 Quadruple 0.398 Widened 2 4471 .40 4439 34 Septuple Quadruple Too diffuse to measure 0.134 Inner pair 0.094 Outer pair 0.258 0.329 Widened 2. Much strength- ened Widened 2. Blends with 4489.3 4527.48 Septuple (0.134 (^0.122 Inner pair 0.232 Outer pair 0.510 Widened 3. No evidence of doubling 4529.65 Quadruple O.I8I 0.258 Blends with 4529.7. Appar- ently little affected 4544.83 Septuple (0.127 0.497 (mean for two pairs) Blends with 4544 .7< Much strengthened. Widened 1-2 4590.11 617 .45 4623.24 Quadruple Quadruple Quadruple 0.207 Not clearly resolved 0.091 0.257 0.311 0.288 Widened i? Slislitly weak- ened. Enhanced line Widened 2. Much strength- ened Widened 2. Much strength- ened 4629.52 4639.54 Quintuple Quadruple (0.136 (0.132 0.163 Too weak to measure 0.457 Widened i. Slightly strength- ened Widened 3. No evidence of doubling Widened 3. No evidence of doubling Widened 4. No evidence of doubling. Fringed to the red 4639.85 4640.12 Quadruple Quadruple 0.138 0.254 Too weak to measure 0.592 4645.37 Triple Not separated 0.244 Widened 4. Probably double 4698.94 Quadruple 0.081 0.293 Widened 2. Much strength- ened Much strengthened but narrow. Widened i ? Widened 2. No evidence of doubling Widened 3, No evidence of doubling Widened 3. No evidence of doubling 4710.34 5016.32 5020.17 Quadruple Quadruple Quadruple Not clearlv resolved 0.204 0.218 0.375 0.426 0.384 5023.02 Quadruple 0.241 0.374 * Theee folnmiw refer to the spark in the magnetic field when oheerved acroM the Hnes of force and along uKBk respectively. 57 i I ARTHUR S. KING TlTASlVM—CotUinued 5025.03 5066.17 5085.51 5109.60 5418.98 5426.47 5460.7a 5481.65 5712.07 5716.67 57ao.57 5903 54 5938.04 5941.98 6064 85 6085.47 6303.98 6312.46 Spark Obierved at Right An^lea to Force Lues Quadruple Quadruple Quadruple Quadruple Quadruple Quadruple Quadruple Quadruple Quadruple Quadruple Ak Spark, Trana- veiae Effect* Quintuple Triple Quadruple Quadruple Triple Quadruple Quadruple Quadruple 0.412 0.283 0.319 0.259 (diffi- cult) 0.292 0.408 0.431 0.429 o 313 0.417 0.630 Not separated Wide and hazy, not clearlv re- solved 0.456 Not separated 0.225 0.487 0.367 AA Spaik, Longitudinal ^Sea* 0.108 0.227 Not measurable Not measurable 0.395 0.532 0.568 0.429 0.550 0.256 i 0.338 o 0.340 0.601 0.612 o.5a7 0.876 0.808 0.493 0.615 daiacter in Sun-Spols Widened ^~4- No evidence of doubbng Blends with 5056.08 Cr. Difficult. Apparently wid- ened 2 Widened 3 Widened 2 Large shift to red in spot Widened 2. Maximum to- ward red side of line. En- hanced line Widened 3. Probably double. Measured separation ao85 Widened 4. Maximum at about center of line Narrow. Fringe to violet making hazy, continuous band as far as X 5481 .45 Widened 4 but blends with J712.10 Fe, Difficult to judge structure. Possibly double Widened 4. Possibly double though line appears of nearly uniform intensity throughout. Apparent com- ponents measiured 0.105 Widened 4. Narrow, sharp ^ maximum in center of line Widened 4. Probably double. Apparent ^components mess- ured 0.094 Widened 4. Double. AX— 0.093 Widened 4. Double. Diffuse center. AX— 0.113 Widened 5. Triple. (0.087 AX- j o (0.092 Widened 4. Not clearly doubled Widened 3. Double. AX-0.087 Widened 3. Double. AX— 0.090 * These columns refer to the apark in the magnetic fidd when ohaer^ed acroas the linea of force and akof them respectively. 58 SI SEPARATION OP SUN-SPOT DOUBLETS S Plate II reproduces the iron spectrum from A 6213 to X 6337 given by the spark in the magnetic field. The upper spectnun shows the doublets obtained when the spark is observed parallel to the lines of force, the so-called 'longitudinal component." The lower was made vnth the light at right angles to the force lines and passing through a Nicol prism tiuned so as to transmit the light vibrating parallel to the field, the "transverse component." The quadruplets thus appear as narrow doublets in the latter spectrum, while the triplets show their middle component single. The purpose of the table is to show to what extent a correspond- ence exists between the structure of lines given by the spark in the magnetic field and by the sim-spots. This may be discussed best by considering each type of- separation in turn. I. The fairly large group of quadruplets in the spark is of special interest when compared with the list of spot doublets. For iron there is practically a complete correspondence for the limited region in the red thus far examined with the Nicol prism. The only appar- ent exception is X 6256.57. This is much widened in the spot but does not appear doubled. In the spark it is an imusual type of quadruplet. The transverse component is clearly separated (see Plate II) while the parts of the longitudinal component are wide and hazy and not clearly resolved. A line of this character, in the weak field of a sun-spot, would have the central portion so filled up that no separation of the parts could be effected. For titanium the correspondence is very dose for lines in the yellow, orange and red, b^inning at about X 5700. The most important exception is X 5903 . 54, which is much widened in spots and probably double. This line shows in the spark the structure of the normal triplet, the transverse component being narrow. Two extreme types of quadruplets are represented in X 5941 . 98 and X 6085 . 47. The former has in the spark the character of the iron line X 6256. 57 mentioned above, the transverse component having a wide separation while the longitudinal is separated but little more and the central portion is so filled up by the diffuse and hazy components that a measurement is difficult. The spot line is of an analogous type, having the center filled up, but the maxima at the sides are sufficiently distinct to enable it to be classed as a doublet. X 6085 . 47 has its transverse component very 59 6 ARTHUR S. KING narrowly separated, with a widely separated longitudinal component. In the spot this line is widened, but not doubled. In the region of shorter wave-length, the agreement is not nearly so good. X 5645.37 shows the same relation in spark and spot as k 5903 . 54. Several weak and doubtful spot doublets in the blue and green are probably narrow quadruplets in the spark, but are not included in the table. The lines in this region which are dear quad- ruplets in the spark, however, have no uniform behavior in the spot. Many are considerably widened (i to 3 on a scale of 5), but do not show doubling, while a few are sUghtly narrowed in the spot. The change in correspondence begins to show itself in the neighborhood of X 5000, and the lack of agreement with the region of greater wave-length points to a masking of the real character of the spot lines, presumably by the same agency which causes other influences, such as temperature, to show a much closer accordance between laboratory and spot results in the less refrangible region of the spectrum. 2. The few lines of iron and titanium occurring as triplets in sim-spots show a complete agreement with the laboratory results. XX 6173.55 and 6302.71 of iron and X 6064.85 of titaniimi are all wide triplets in the spark, the middle component remaining sharp. The latest measurements of the best photographs show that X 6302.71 is not asymmetrical, and indicate that the appearance of asymmetry previously noted was due to disturbing influences in both spot and spark spectra. 3. The few titanium lines appearing as quintuplets and of still higher separation in the region studied are of two types. XX 4464.62 and 5720.57 appear as triplets when observed along the lines of force in the laboratory and as quintuplets across the lines of force, two more components being added on the outside when viewed at right angles. Such a line will then have a strong central component at whatever angle to the force lines the light may be taken. X 5720.57 is favorably located for comparison with the sun-spot line. In the spot the line is very much widened and has a narrow and unusually strong maximum in the center. The structure of the spark line should give this appearance, the quintuplet structure giving a strong widening, while the central maximum is accounted for. In X 4464.62 the strength of the central component is probably partly responsible 60 SEPARATION OF SUN-SPOT DOUBLETS 7 for the apparent narrowing of a rather faint line, the sides being weak and indistinct by contrast with the center. The other type has one or more pairs of rather widely separated components when viewed along the lines of force; while across the lines of force a dose triplet is seen in the middle of the structure. Lines of this type are XX 4471.40, 4527.48, 4544.83, and 4629.52, and by referring to the table these are seen to be both widened and strength- ened in sim-spots, in no case doubled. This again accords well with the spark structure, the widely separated components giving a broad line, while the central maximum is not so conspicuous as in the first type. 4. One strong unseparated line stands out with great distinctness in the titanium spectrum. This 15^5714.12, which appears equally narrow in both positions of the Nicol. This line is remarkable for its narrowness in the spot spectrum. We have thus a good deal of evidence that light coming at an angle to the lines of magnetic force plays a large part in giving the sun-spot lines the structure which is observed. The presence of trip- lets in the spot spectrum is fully explained. The fact that the spot doublets occiu: very generally as quadruplets in the laboratory when observed at right angles to the lines of force obviates the necessity of assuming that doublets in the spot are due to the action of the field along the lines of force alone, since in either case the central com- ponent would be absent for these lines. The two inner components of the quadruplets are seldom widely separated, even in the fairly strong laboratory fields; so that what we probably have in the spot lines is a weakening of the center of the line due to a narrow separation of the transverse component, throwing the light into maxima at each side. The transverse effect becomes weaker as the direction of the light becomes more nearly parallel to that of the lines of force, giving a more distinct separation of the components of the doublet. This dependence oi the relative intensity of the components upon the angle of the field is unfavorable to quantitative relations between the measured separations of spark and spot lines. It is quite probable that some lines are produced more strongly than others in certain parts of the spot, on accoimt of differences in temperature and in other physical conditions. This 61 8 ARTHUR S. KING would involve a difference in field strength and also of the angle to the lines of force at which the light comes to us, opening the way for many differences in the appearance of the lines in the spot and in the laboratory sources. For example, light coming almost parallel to the force lines of the solar magnetic field would give a strong pre- ponderance of circularly polarized light. Another part of the spot, still in the field of view projected on the slit, but differing either in location or in level or both, and therefore in a different part of the solar magnetic field, might send light at a larger angle to the lines of force, giving to some extent the transverse effect and producing the central line of the triplets. This may seem to require somewhat artificial conditions, but a structure of the spot allowing for this would appear necessary to account for the relative intensity (about | to i) of the middle and side components of spot triplets such as >U 6173 and 6302. This would require a mixture of the longitudinal and transverse effect corresponding to an angle not yet determined exactly, but greater than 30^ to the lines of force and much less than 90°. Many of these doubtful points will, it is hoped, be cleared up as the investigation is extended with improved facilities. Mount Wilson Solak Observatory December 5, 1908 6a t X % ^ 6-Vo, v3ro contributions from the mount wilson solar Observatory NO. 3S A THE RELATIVE INTENSITIES OP THE YELLOW, ORANGE Alrt) RED UNES OP CALCIUM IN ELECTRIC PURNACE SPECTRA BY ARTHUR S. KING / ( Preprinted from the Astfophysical Journal^ Vol. XXIX, April 1909 ,<■ Contributions from the Mount Wilson Solar Observatory, No. 35 I^eprinted from the Aslropkysical Journals Vol. XXDC (April) 1909 THE RELATIVE INTENSITIES OF THE YELLOW, ORANGE AND RED LINES OF CALCIUM IN ELECTRIC FURNACE SPECTRA By ARTHUR S. KING The large electric furnace in the Pasadena laboratory' has recently been used by the writer to study the effects of different temperatures on the calcium spectrum in the region of greater wave-length, specia attention being paid to some twenty lines occurring from X 5857 to A 6718, which were affected very differently by changes in the furnace temperatures. All of the prominent arc lines in this region are given by the furnace, and the experiments were arranged to bring out not only the effects of varying the temperature, but also of increasing the quantity of vapor, to observe whether any lines in this region show a behavior similar to that of the " flame line" X 4227, which was shown in a former investigation' to respond but slightly to change of temper- ature, while enormous widening was brought about by a large increase in the amount of calcium vapor present. The photographs were made with the vertical Littrow spectrograph w^ith an objective of 13 feet (4 m) focal length, the first order of a 5-inch (13 cm) plane grating being used. The plates were Seed's " 27," sensitized by the Wallace three-dye process.^ The furnace was charged with a few pieces of clean metallic calcium, a large or small amount being used according to the vapor-density desired. The tube was heated in vacuum, first with an e.m.f. of 20 volts which rapidly raised the temperature, reaching in about five minutes a temperature of approximately 2800° C. as measured by a Wanner pyrometer. An exposure was made as the temperature rose, then others of one- half to one minute at the maximum temperature. A vigorous vapori- zation of the carbon was then taking place. The current was broken ' Contributions from the Mount Wilson Solar Observatory, No. 38; Astropkysical Journal, a 8, 300, 1908. > Contributions from the Mount Wilson Solar Observatory, No. 32; Astrophysical Journal, a8, 389, 1908. 3 Astrophysical Journal, 26, 299, 1907. 63] I 2 ARTHUR 5. KING and IS volts substituted, after which one or two low-temperature expo- sures were made, fifteen to twenty times the length of the exposure now being required to produce a photograph of the same average strength as that at the highest temperature. No dose value for this temperature can be given, as it was somewhat variable, the heating of the jacket- ing running up the temperature so that is was necessary frequently to break the current for a few seconds. The average temperature seemed to be about 2200° C, probably never rising above 2400° C. The influence of vapor-density was tested by using very different amounts of calcium in the furnace tube. These were given (i) by the empty tube, the graphite of which contains enough calcium to show the stronger lines distinctly; (2) a small amount (about o.i gm) of metallic calcium; (3) a large amount (2 to 3 gms) of the metal. It may be noted here that chemical action was present in these experiments to a greater degree than is the case when a substance not readily oxidizable is used in the furnace. The bright surfaces of the calcium became dulled while it was being placed in the furnace tube and further time was required to close the chamber and pimip out the air. The result was that brilliant red bands appeared dur- ing the first minute of the heating. These soon died out for the most part, but some of them persisted through several exposures, especially when a large quantity of calcium was used. These weakened rapidly with time, and the strength of the calcium lines appeared to be quite independent of the existence of the bands, the lines being produced with equal facility whether the bands were strong or barely visible. The following table includes the lines from X 5582 to X 6718, with their intensities in the furnace spectra at high and low temperatures for the three conditions of vapor density given respectively by a large amount of metallic calcium, a small amount, and by the empty tube. The estimates of intensity were made by comparison with a photo- graphic scale in the manner described in a previous paper,' modifying the estimates when necessary according to the character of the lines. The spectra given by high vapor-density afforded the most complete material for comparison, on account of the average density of the two photographs being very nearly the same; but some allowance was > Contributions from the Mount Wilson Sotar Observatory^ No. 28; Astrophysical Journal^ a8, 300, 1908. 64 K ii 5.ji CALCIUM LINES IN ELECTRIC FURNACE 3 necessary for the lines at the higher temperature being widened and difiFuse when much vapor was present, instead of sharp and black. In the spectra with less calcium present, the spectrum for the lower temperature was as a whole weaker than for the higher temperature, so that a factor of 2 is needed to bring the low-temperature spectra (in the fifth and seventh columns) to an average intensity comparable with that at the higher vapor-density. The intensities are given in the table without reduction. High VAPOK-DEMsmr Medidx Vapos-Demsitt Low VAPOR-DBNSiry A High Temp. Low Tempu High Temp. Low Temp. High Temp. Low Temp. 5583.20 2 I 2 trace * I 5588.98 4 3 4 2 4 3 5590 -34 2 trace I • • * trace 5594.69 3 2 3 I 3 2 5598.71 3 2 3 I * 2 5601.50 2 I I • • * I 5603.08 2 I I • * * I 5857-67 5 2 3 trace 5 I 5867.78 2 I 2 I I • • 6102.94 9 8 9 6 9 6 6132.43 10 18 12 15 16 10 6161.50 I • • trace • • ■ • m • 6162.39 8 24 14 22 20 15 6163.97 2 I I ■ • I ■ ■ 6166.65 3 I 2 trace 2 • • 6169.25 4 2 3 trace 4 2 6169.78 6 3 7 I 6 3 6439.29 16 18 18 9 20 10 6450.03 10 6 8 2 6 3 6455 82 4 • • 3 • m 2 m • 6462.78 14 IS 16 7 15 7 6471.88 8 4 5 I 4 2 6494.00 12 9 10 3 10 4 649988 7 2 3 trace 2 I 6573.03 5 18 5 10 8 II 6708.18 20 80 24 60 30 70 6717.94 3 2 I • • 2 • • * Concealed by strong carbon fluting in this photograph. Description of plate. — Plate III reproduces the region from X 5890 to X 6708 for different temperatures and diflferent quantities of vapor Photograph No. i gives the spectrum from calcium in the carbon arc a large quantity of the metal being present. Nos. 2, 3, and 4 are furnace spectra with much calcium in the tube, 20 volts being used on the tube for No. 2 with an exposure of one-half minute when the tube was very hot. Nos. 3 and 4 were taken with 15 volts on the tube and exposures 65 4 ARTHUR 5. KING of 5 and lo minutes respectively after No. 2 had been made, the current being broken several times during No. 4 to keep the tempera- ture down. No. 4 is most comparable in average intensity with No. 2, though No. 3 is useful to show the relative strength of some of the stronger lines at the lower temperature. Nos. 5, 6, 7, and 8 are furnace spectra with a small quantity of calcium in the tube. Nos. 5, 6, and 7 were taken with 20 volts on the tube and successive exposures of four, three-quarters, and one-half minute respectively as the temperature rose, there being no break of the current between exposures. No. 7 best represents the condition for high and nearly constant temperature, while Nos. 5 and 6 are of interest as showing the weakening of the bands as the vaporization progressed in the vacumn, it being possible to keep the lines of nearly the same intensity in the three photographs, by proper timing of exposures. No. 8 is a low-temperature spectrum taken with 15 volts on the tube and an exposure of seven minutes inmiediately after No. 7. A final arc photograph (No. 9) is added below for comparison, the amount of calcium and exposure time being so adjusted as to give the prominent Unes with very weak bands. Discussion. — Referring to the table, it is obvious at once that the calcium lines in this region differ greatly in their response to the stimu- lus accompanying increased temperature. The low-temperature lines may be at once selected from the first two columns of the table or by comparing any pair of photographs at high and low tanperatures reproduced in Plate III. These lines strengthen but slowly with rising temperature, while a number of lines barely visible at the lower temperature rapidly attain considerable strength with higher heating of the tube. Several sets of lines may be selected which show this effect in a striking manner, for example, X 6122 and A 6162 as compared with X 6103; X 6439 ^"^^ ^ 6463 as compared with X 6450 and X 6456; X 6573 and X 6708 as compared with X 6718, and a number of other notable contrasts. X 6103 must, however, be classed among the low-temperature lines. The object of using different quantities of calcium in the furnace tube was to see how large a part vapor-density may be expected to take in determining the relative intensity of the spectrum lines, calcium being in many ways a favorable substance for a test of this kind. Comparing the first and third columns of the table, these spectra being 66 I I CALCIUM UNES IN ELECTRIC FURNACE $ taken at approximately the same temperature and with very different quantities of vapor, a few of the weaker lines show slightly stronger with higher vapor density, the most decided being A 6471.88 and i 6499.88. An interesting difference is seen between A 6122 and A 6162, their intensities being in the ratio 10 to 8 at the higher vapor density and 12 to 14 at the lower. It holds in all of the spectra, how- ever, that A 6122 strengthens with rising temperature more rapidly than A 6162, so that if the tube were slightly hotter for the spectrum given in the first column the diflFerence between the two lines should be in the direction observed. The same cause would account for the behavior of X 6471 .88 and X 6499.88, both of these being high- temperature lines and rapidly strengthened as the temperature rises. The line which we should expect to be most afifected by vapor- density is the strong red line X 6708, which is comparable in many ways to the " flame line" X 4227, appearing strong at low temperatures with a mere trace of calcium present. It does not readily reverse, however (the widened appearance is due in part to poor focus at the end of the plate), and shows no decided change in appearance or intensity when the quantities of vapor are greatly different. An examination of plates taken in this and the former investigation, covering the spectrum as far as A 3700, shows no lines except X 4227 in which change of vapor-density has a decided effect. The con- clusion to be drawn from the calcium spectrum is that while increase of vapor-density makes aU of the lines wider and more diffuse, it does not appear promising as a general cause in producing changes of relative intensity among lines when temperature and other condi- tions remain unchanged; and these latter can be held constant for different vapor-densities much more nearly in the furnace than is possible in the arc. It appears probable that X 4227 and some of the niost decided "flame lines" in other spectra are in a class by them- selves as regards dependence on vapor-density. However, a study imdex higher dispersion of the widening produced by increased vapor- rjgP^ity will be necessary for a full discussion of this point. The band spectrum in this region is of interest for the present paper chiefly as an indication of the degree to which chemical action is present. The bands which appear in the furnace photographs occur, ^tti one exception, also when metallic calcium is vaporized in the 67 6 ARTHUR S. KING carbon arc burning in air and are presumably due to the oxide unless possibly they are given by the metal itself. As has been noted, the calciimi became oxidized to some degree before the air could be removed from the furnace chamber. Photographs Nos. 5 to 8 of Plate III show the weakening of the band spectrum as the experiment pro- gressed and indicate that this initial state of oxidation gave rise to the the bands. The increase of temperature during exposures 5, 6, and 7 might act to weaken the bands, but in that case No. 8, taken at a a lower temperature comparable to the beginning of No. 5 when the bands were most brilliant, should restore their intensity to some degree, which is not the case. The same argument makes it doubtful as to whether they can belong to the metal; since in that case they should show a behavior similar to that of the low-temperature lines (A 6708, for example) instead of a steady fading away as time went on. When a larger amount of calcium was used, the bands were much more persistent, as would be expected from the increased opportunity for chemical actipn. In No. 2 the stronger bands are seen reversed, while they are bright and fairly distinct in Nos. 3 and 4. The band referred to, which does not appear in the spectrum of the arc burning in air, is that with prominent heads at A 6389.3 and X 6382.2, shaded toward the violet. They show distinctly in Nos. 2 and 5, especially in the latter, and are faintly visible on the negative in the low-temperature photographs. This is the band measured by Olmsted* and found especially strong in the calcium arc burning in hydrogen. The band is not usually perceptible in the calcimn arc in air or when a plentiful supply of oxygen is present. Hydrogen or water vapor suppUed to the arc with partial exclusion of oxygen bring it out strongly. In the furnace spectra it is favored, like the other bands, by the initial conditions when the furnace is heated up. It weakens with time and is not restored by lower temperature, so that it seems unlikely that this band belongs to the metal; while its behavior in the arc, as has been noted, is differ ent from that of the other furnace bands. If the band at X 6382-6389 is due to a hydrogen compound, this would most probably arise from moisture inside the steel chamber, which would be vaporized at the X Contribuiions from the Mount Wilson Solar Observatory, No. 21; Aslrophysical Journal, 27, 66, 1908. 68 CALCIUM UNES IN ELECTRIC FURNACE J first heating, and would then be rapidly pumped out, as the pump is usually kept working during the operation of the furnace. It is hoped soon to make further tests as to the origin of the band. Comparison with suthspot lines. — ^The sensitiveness of the calcium lines to temperature changes makes them favorable for use in testing the hypothesis that temperature is one of the chief agencies in causing the differences between the solar spectnmi and that of sun-spots. The relation holds, quite as generally as in any spectrum examined, that all of the low-temperature lines are much strengthened in sxm-spots, usually in the ratio of about 2 to i in photo- graphs where the continuous ground is of the same density for both spot and solar spectra; while the two most conspicuous low-tempera- ture lines, X 6573 and X 6708, are intensified in sun-spots in the ratio 15 to I and 5 to coco (on the Rowland scale) respectively. However, some condition exists on the sim which causes all of the calcium lines to be more or less strengthened in the spot and it frequently happens that some of the high-temperature lines, as for example XX 5858, 6167, 6456, 6500, are considerably stronger in the spot than in the sun. There is a vital diflference, however, between the character of these lines and that of the pronounced low-temperature lines, such as XX 6103, 6122, 6162, 6439, 1^ ^^^^ th^ latter are very much widened and "winged" in the spot, the formation of wings being very slight for the group of high-temperature Unes. So much remains to be settled in regard to the causes of widening in solar lines, the matter being stiU further complicated by the existence of a magnetic field in sun-spots, which would cause lines to be widened differently according to the type of magnetic separation, that for the purposes of the present paper the matter may best be left with the statement just given of the general relation between the low- and high-temperature lines. An intensive study will soon be made of a few of the calcium lines under higher dispersion, to determine by a comparison of various laboratory conditions what causes are chiefly instrumental in the observed widening of these lines in solar spectra. It may be well to state what is meant by the constant use in my electric furnace papers of expressions concerning the dependence of spectra upon "temperature," that necessary brevity of statement in the use of this word may not be mistaken for looseness. Nothing 69 8 ARTHUR S. KING in the electric furnace experiments can claim to throw any light on the ultimate mechanism of radiation. We are, however, dealing here with temperature as the primary or rather the initial exciting cause; though the experiments leave open the possibility that other agencies, results of the high temperature, supply the stimulus finally needed to produce light. With the low voltage employed on the furnace tube, it is scarcely conceivable that the electric current pro- duces any condition that would not be present if the tube were heated by a Same applied to the outside or were imbedded in a mass of hot material, the ease of producing a high temperature being the sole point in favor of the use of the electric current. All of the consequences of the electron theory may hold for the radiation in the furnace, provided we admit that the temperature supplies the necessary energy and controls its amount. This position of temperature as the primary excitation is the essential point of contrast between the furnace on the one hand and the arc and the flame on the other. In the two latter sources, high temperature is the result, in the one case of the electrical action, in the other case of the chemical processes of combustion. Mount Wilson Solar Observatory January 1909 70 J ^ • ) - r" ! • .T ^oo^i'-LO.Ci'o ®antf 0tr SttBtttntiiiti of Vlaslfbiglittt Contributions from the Mount Wilson solar Observatory NO. 86 A .** •/ w THE 60-INCH REFLECTOR OF THE MOUNT WILSON SOLAR observatory BY G. W. MTCHEY Preprinted from the AUrophyncal Joumaiy Vol. XXIX, April 1 909 L-'^ "»'' •* ■» \ I ' I OF THE MOUNT WILSON ERVATORY UTCHEY jnt Wilson Solar Observatory was 1 the night of December 13, 190S, was secured with it on December the telescope and its steel dome dieted; the most important of these the small refrigerating apparatus hout the day at the expected night r is designed to be used in four as a Newtonian, for direct photog- urier, and for spectroscopic work Newtonian focus; in this use the see Fig. r). Second, as a Casse- ith the double-slide plate-carrier; 3 G. W. RITCHEY in this use the equivalent focal length is approximately lOO feet (30.5 m), and the enlarged image is formed at the north side of the tube, near its lower end (see Fig. 2). Third, as a Casse^rainian for spectroscopic work with a large spectrograph, of the type of the Bruce spectrograph of the Yerkes Observatory, attached to the north side of the strong cast-iron part of the tube, near its lower end; in Fio. 3 Fio. 4 thb use the equivalent focal length is approximately 80 feet (24.4m) (see Fig. 3). And fourth, as a Cassegrainian-CotM^/ for spectroscopic work with a very large spectrograph mounted on stationary piers in an underground constant-tem[>erature pit; in this use the equivalent focal length is approximately 150 feet (45.5 m) (see Fig. 4.) THE OmCAL PAETS For the above uses of the telescope six optical mirrors are required, all of which have been completed in our shop: the 60-inch par- aboloidal mirror of 399 (7.6 m) inches focal length, which in its finished condition is yf inches (19 . 4 cm) thick at the edge, 6| inches (17.5 cm) thick at the center, and wei^s 1900 lbs. (865 kilos); the Newtonian plane mirror, which is of elliptical outline, is 19} inches (50.3 cm) long, 14) inches (36.8 cm) wide, and 3I- inches (7.9 cm) thick; the Coudt plane mirror, which is also elliptical in THE MOUNT WILSON 60-INCH REFLECTOR 3 outline, is 22 J inches (56.5 cm) long, 12 J inches (31.8 cm) wide, and 3} inches (9 . 2 cm) thick; and three convex hyperboloidal mirrors, with diameters respectively of 16 inches (40.6 cm), ifrj inches (42.6 cm), and 17} inches (44.5 cm), and giving equivalent focal lengths, as before stated, of approximately 80 feet (24.4 m), 100 feet (30.5 m), and 150 feet (45.7 m) respectively; each of these mirrors is about 3 inches (7.6 cm) thick. All of these mirrors are polished approximately flat on the back, and when in use in the telescope are silvered on the back as well as on the face, in order that the effect of temperature change may be symmetrical on front and back. The methods used in grinding, polishing, and testing the mirrors are practically the same as those described in my paper on ''The Modem Reflecting Telescope," published by the Smithsonian Institution in 1904. The methods used in testing the optical sur- faces are also described in the Astrophysical Journal, 19, 53, 1904. As these methods have now been applied and thoroughly tested in the case of a reflecting telescope of the largest size with satisfactory results, the following remarks upon the optical work may be of interest. Of much practical importance in its bearing upon the making of very large optical surfaces is the fact that, in the case of the 60-inch glass, grinding and polishing tools of only about one-fourth the area of the glass were used with entire success in excavating the large concave and in fine grinding and polishing; a full-size, flat grinding tool was used only in the preliminary work of securing a perfect surface of revolution. A circular grinding tool of cast-iron, 31} inches (80 cm) in diameter, was used in all of the fine grinding of the large concave surface. In polishing this concave, and in bringing it to an optically perfect spherical surface preparatory to parabolizing, a 90*^ sector-shaped polishing tool of exactly one-fourth the area of the large glass was used with the best results. In par- abolizing, a circular polishing tool 20 inches (50.8 cm) in diameter was exclusively used in securing the necessary change of cxirvature from center to edge of the glass; in addition to this the 90^ sector tool, used with long diametrical and chordal strokes, was found to be of great value in smoothing out the paraboloidal surface. With these two figuring tools alone, used with the machine, a very close approximation. to a true paraboloid was secured. The figuring was 73 4 G. W. RITCHEY completed with much smaller tools used by hand to soften down several slight high zones. In figuring the large paraboloid, one modification only was foimd desirable in the polishing machine described in my Smithsonian paper. The two cranks which give the motion to the polishing tools were remade in such a way that their throw or stroke can now be altered at will while the machine is running: The optician is thus enabled to change the position and stroke of the tool with a perfectly smooth progression while parabolizing; these changes are actually made at the end of each revolution of the glass, and a very great improvement in the smoothness of curvature of the parabo- loid is at once apparent. In the early stages of figuring the large paraboloid, testing was done at the center of curvature, by measuring the radius of cxirvature of the successive zones; in the final stages, however, all tests were made at the focus of the paraboloid, with the aid of a 36-inch (91 . 4 cm) collimating plane mirror of the finest figm^, which was made in our optical shop expressly for the purpose of testing the large paraboloid and the three smaller hyperboloidal mirrors described above. This 36-inch plane mirror was mounted on edge on an iron carriage sliding on massive iron ways carefully finished straight, and could be moved horizontally by means of a long screw; in this manner it could be readily placed so as to show, at the testing knife-edge at the focus of the 60-inch paraboloid, any 36-inch circle of this paraboloid. This test is a most rigorous and satisfactory one, enabling the optician to see, and to determine the character of slight zonal errors which cannot be detected by the test at the center of curvature. The three hyperboloidal mirrors were tested in a similar manner, in conjunction with the 36-inch plane mirror. While a collimating plane mirror of the full size of the paraboloid would be desirable and convenient, my experience has shown that a plane mirror with a diameter three-fifths that of the paraboloid, used as above described, gives excellent results. A rough disk of glass has accordingly been ordered for a 60-inch plane mirror for testing the loo-inch (254 cm) Hooker glass. THE MOUNTING In all essential features the design for a 60-inch moimting de- scribed in my Smithsonian paper has been carried out. The general 74 THE MOUNT WILSON do-INCH REFLECTOR 5 character of the mounting will be seen by reference to Plates IV and V; the first of these shows the mounting as it appeared in our erecting shop in Pasadena, the second as it appears when finally set up in its dome on Mount Wilson, The very large parts of the mounting, including the base, the polar axis, the mercury float and trough, and the tube, were made by the Union Iron Works Company of San Francisco; a large amount of final machining and finishing of these parts was done by us after their arrival at Pasadena. In addition, the construction of all of the smaller and more refined parts of the mounting, including the driving-clock and its connections, the electric quick- and slow- motion mechanism, the lever-support system of the large mirror, the cells and their supports for the five small mirrors, the automatic- rotation mechanism for the Coud^plajie mirror, the large double- slide plate-carrier, the graduated circles, and the cutting and grinding of the lo-foot (3.5 m) worm-gear on the polar axis, was done at our shop in Pasadena. The cast-iron base is 15 feet (4.57 m) long, 7 feet (2.13 m) wide, 18 J inches (47 cm) deep, and weighs 14,000 lbs. (6350 kilos). The north and south columns, forming the bearings of the polar axis, weigh respectively 9500 lbs. (4275 kilos), and 2000 lbs. (907 kilos). The polar axis is a hollow forging of nickel-steel, hydraulic-forged by the Bethlehem Steel Company, and is turned and ground all over; it is 15 feet (4.6 m) long, varies from 1 5 to 18 inches (38 . i to 45 . 7 cm) in diameter, and weighs 9200 lbs. (4140 kilos). At its upper end is a head or flange 4J feet (i . 37 m) in diameter and 6 inches (15.2 cm) thick. To the lower side of this flange is bolted the float, which is a very rigid hollow disk of steel boiler plate, 10 feet (3 . 05 m) in diameter and 2 feet (61 cm) deep or thick, weighing 8600 lbs. (3900 kilos). To the upper side of the flange of the polar axis is bolted the fork, between the great arms of which the tube swings in declination on nickel-steel trunnions 7 inches (17.8 cm) in diameter. The fork is of cast-iron, of hollow box-section, is 9 feet (2.7 m) across in extreme width, and weighs 10,400 lbs. (47 11 kilos). Twelve nickel-steel bolts, 2 J inches (6.4 cm) in diameter and 3 feet (91.4 cm) long, pass through reamed holes through the base of the fork, through the flange of the polar axis, and through the cast-iron center or hub of the float, thus clamping these massive parts together with 75 6 G. W. RITCHEY extreme strength and rigidity at a region of the mounting where the greatest tendency to flexure occurs. The float dips in a cast-iron trough which is machined to nearly fit the float, leaving a space of only one-eighth of an inch all around; this space is filled with 650 lbs. (295 kilos) of mercury. The inmiersed part of the float gives a displacement of about 50 cubic feet (1.4 cu. m) of mercury, thus carrying 21 J tons (19,479 kilos) of the moving parts of the telescope in the fluid, and relieving 95 per cent, of the weight on the large bearings of the polar axis. The mounting is so designed that the center of weight of the moving parts is vertically above the center of flotation. The large worm-gear for the diurnal rotation of the telescope is 10 feet (3.05 m) in diameter and has 1080 teeth. While being cut, these teeth were spaced with the utmost care with the aid of a 36-inch (91 .44 cm) Warner and Swasey graduated circle of the finest quality. The teeth were then hobbed (with a hob of special design by which the accuracy cannot be lost) and were then ground with hone powder of finer and finer grades, with oil, and were finally polished with rouge and oil. This treatment not only eliminates any small irregu- larities of spacing, but leaves the teeth exquisitely smooth. The driving-clock is in many respects a copy of the driving-dock of the 40-inch Yerkes refractor, built by Warner and Swasey. I have introduced one important modification, however, as follows: the clock-governor is driven by a weight through a spur-gear train, as usual, but the motion of the governor is conmiimicated to the telescope through the medium of a worm and worm-gear, which are ground with the utmost care to eliminate periodic errors. This worm-gear of 80 teeth, together with the large worm-gear of 1080 teeth on the polar axis, gives the entire reduction from the clock- governor, rotating once in a second, to the polar axis, which rotates once in 24 hours. Furthermore, the two worm-gears named are the only gears used in communicating the rotation of the clock-governor to the polar axis, the driving-clock being so placed that all spur- gears and bevel-gears are dispensed with. To relieve friction and wear on the polished teeth of the large worm-gear, and on the clock and clock-connections, the following simple expedient is used. A small wire-rope passes over a grooved wheel keyed to the polar axis, runs over two grooved pulleys on th^ 76 ^ THE MOUNT WILSON do-INCH REFLECTOR 7 west side of the telescope base, and is loaded with about 100 lbs, of iron weights hanging vertically on the west side of the pier. With this assistance, only about two pounds' pressure on the teeth of the lo-foot worm-gear is required to rotate the moving parts toward the west. As a further means of keeping the teeth of the large worm-gear continually in the finest condition, a small motor is provided by means of which the gear and its worm can be repolished at any time. In practice this is done several times each week; the worm-shaft is disconnected from the driving-clock by simply removing two small screws, fine graphite and oil are supplied as a lubricant, and the large worm-gear and its worm are run together for an hour. As a result of the mercury flotation and the care which has been given in finishing the driving-clock, the clock-connections, and the large worm-gear, the great telescope follows the stars with exquisite smoothness and accuracy, despite the fact that the moving parts weigh nearly 23 tons. The telescope is provided with electric quick and slow motions. The former give a speed of 30 degrees per minute of time, in both right ascension and declination. The latter are arranged to give two speeds, one of six minutes of arc in a minute of time, for ordinary fine setting, and a slower one of one-half minute of arc in a minute of time for guiding with the spectrographs. The electric wiring is so arranged that the slow motions can be operated from several convenient points. The octagonal skeleton tube is worthy of description because of its extraordinary rigidity. It consists of eight conical tubes of J inch (3.2 mm) sheet steel, about 15 feet (4.6 m) long, tapering from 5 inches (12.7 cm) diameter at their lower ends to 3 inches (7.6 cm) diameter at their upper ends; each tube is made of two parts riveted together, with two flanges 135° apart extending their whole length; to these flanges are riveted the diagonal braces; three rigid rings connect the eight tubes together. Any one of four interchangeable extensions of the tube, called "cages," of similar construction, can be connected to the upper end of this permanent part of the skeleton tube; one of these carries the Newtonian plane mirror and its accesr sories; this is shown in Plate VI; each of the others carries one of the small hj^erboloidal mirrors, with its cell, cell-support, and small 77 8 G. W. RITCHEY electric motor for focusing. A simple and effective machine, called the "cage-lift," which is suspended from the framework of the dome, enables the observer or assistant to interchange the cages quickly and safely. The Newtonian "cage" or tube-extension can be connected to the tube in four different positions 90^ apart, that is, with the diagonal plane mirror and the double-slide plate-carrier facing either north, east, south, or west. The cage-lift is so designed that the tube- extension can be rotated to the position desired while suspended in it, before being attached to the tube. The double-slide plate-carrier is most carefully designed, and is much more elaborate than those which I have made and used in the past. It is so planned that it will take either 5X7 inch (12.7X 17.8 cm) or 6^X8^ inch (16.5X21.6 cm) photographic plates. This and other features of the design allow a very large range of movement of the guiding eyepiece, for choosing the most suitable guiding-star available — a matter of the utmost importance in using this attachment. Provision is made for altering the plane of the photographic plate when desired, during the exposure, without danger of relative rotation of the field and plate. Mention should also be made of the lever-support system of the 60-inch mirror. The system fully described in my Smithsonian paper has been used without modification. The mirror is " floated " so that no flexure occurs sufficiently large to be detected by optical tests; and, in addition, the position of the mirror is defined so per- fectly with reference to its tube that no wandering or jumping of the star-images, due to the slipping of the mirror in its cell, can be detected in the guiding eyepiece. THE STEEL BUILDING AND DOUE The building which supports the dome is entirely of light steel construction. Twenty columns, each 22 feet (6.7 m) high, form the comers of the 20-sided equilateral polygon. These colunms support twenty horizontal box-girders, which carry the double track upon which the dome revolves. The building has two sheet-metal walls; the inner one is of ^ inch galvanized sheet steel, and is planned to be air-tight; the outer wall, two feet distant from the other, is of light galvanized sheet steel, 78 THE MOUNT WILSON do-INCH REFLECTOR g and serves merely as a sun-protection. A free circulation of air is allowed between the two; both are painted white. Sixteen sheet- metal windows, closing air-tight against heavy rubber packing, are easily accessible from the lower floor, and assist in ventilating the building and dome quickly when desired. The ground floor of the building is of cement; on this floor are the dark rooms and the electric machinery for revolving the dome. Nineteen feet above this is the operating floor, of thin checkered steel plate supported by light steel columns and I-beams. From this floor are operated the dome-drive machinery, the dome-shutter, the wind-screen, the quick and slow motions of the telescope, the right ascension and declination clamps, etc.; from this floor also the right ascension and declination circles are read. On this floor are arranged the silvering carriage (which is necessary when removing the large mirror from the tube, and when silvering it) and the inter- changeable ends of the tube which are not in use on the telescope. In this floor are twelve large trap-doors, 3X7J feet (0.9X2.3 m) in size, which can be quickly opened to assist in ventilating the building and dome. The dome is 58 feet (17,7 m) in diameter, and is of light steel construction with sheet-metal covering, coated inside with granu- lated cork, to prevent dripping from condensation of moisture, and painted white outside. It revolves on double tracks which are machined true, the dome wheels being all double, conical, and fur- nished with the best Hess-Bright ball-bearings. The dome moves with great smoothness. Two motors are used in turning it: one of three horse-power, which is so geared that the dome makes one complete revolution in six minutes; and a one-horse-power motor, which drives a variable-speed machine, by which the speed of the dome can be changed at will (by simply turning a hand-wheel) to any point between one revolution in one hour and one revolution in twenty-five hours; this allows the observer, when working on the observing platform attached to the dome, to be moved with exactiy the right speed to compensate for the horizontal component of the motion of the telescope. The dome-shutter is extremely large, having a dear opening 16 feet (4.9 m) wide. Instead of opening horizontally in halves, the shutter runs back over the dome, as shown in Plate VII; it is opened 79 lo G. W. RITCHEY and closed by a six-horse-power electric motor. A large metal door, below the shutter opening, 1 7 feet (5.2 m) long by 8 feet (2 . 4 m) high, turning outward on hinges at its bottom, can be opened when it is necessary to observe objects near the horizon. A light metal observing platform, 17 feet (5.2 m) long by 9 feet (2.7 m) wide, travels up and down the curve of the shutter opening, by means of a three-horse-power electric motor. This platform can be operated either from the operating floor or from the platform itself, by simply pushing a button; it is so designed that it auto- matically remains horizontal in all positions. In addition to its vertical movement, and its horizontal movement with the dome, the platform can be moved about 30 inches (76.2 cm) radially with. respect to the vertical axis of the dome. This combination of motions enables the observer to reach the upper end of the tube, and to work with the telescope as a Newtonian, with the utmost convenience, in. most positions of the telescope. A wind-screen 17 feet (5.2 m) wide by 35 feet (10.7 m) long is provided, which can be quickly raised and lowered in the shutter opening by suitable mechanism; it is made of heavy black canvas supported on large steel tubes with rollers at their ends. This protects the telescope from wind, and also from lights in the valley. As will be seen in Plate VII, the exterior of the dome is covered with a strong frame-work of steel pipe. This will be covered during the spring, summer, and fall with gores of white canvas, laced on, at a distance of about two feet from the sheet-steel covering of the dome; provision is made for ample circulation of air beneath the canvas. This, together with the white outer wall of the building below, affords a complete sun-protection for the building and dome. Furthermore, the entire building and dome are planned to close air-tight. For this purpose a frictionless water-seal is provided at the junction of the building with the dome; and all outside doors and windows dose tightly against heavy rubber packing. The dome- shutter also is lined all around with air-tight cushions which can be pressed tightly in place by means of two levers operating a series of toggle-joints. In the early morning, after a night's work, the dome and building will be closed, not to be opened until after sunset, and thus, a great volume of 120,000 cubic feet (3360 cu.m) of cool night air will be shut in air-tight. It is believed that this provision, 80 THE MOUNT WILSON 60-INCH REFLECTOR II together with the complete sun-protection of the dome and building, will reduce the rise of temperature within the structure during the day to a very few degrees. This protection from daily temperature changes shoidd be sufficient for the telescope mounting and for the smaller mirrors. To further protect the large mirror during the day, a small refrigerating plant will be installed within a few months, which will supply constant-temperature air, at the expected night- temperature, dradating through a jacket inclosing the entire lower end of the telescope tube. The necessity for this protection of the laige mirror from the daily rise and fall of temperature, to preserve the finest optical figure, was fully demonstrated by a long series of experiments in the optical shop with this mirror after its completion. It will be seen also that a serious effort has been made in the design of the dome and building to eliminate the so-called dome- and building-effect, that is, the local effect upon atmospheric defini- tion caused by heat-radiation from, or air currents in, the dome and building, which is often a most serious detriment to the successfid performance of large telescopes. An hour after sunset the 16 ventilat- ing windows near the lower floor of the building, the 12 trap-doors in the operating floor, and the great dome-shutter, the latter having an opening 16X45 feet (4.9x13.7 m) in size, are all opened, and the light metal colimims, girders, and walls, inside and outside, quickly assume the temperature of the night air. The ventilating doors and windows are then closed, to prevent draughts. The very fine definition which we have already had on Mount Wilson with the 60-inch aperture, even on winter nights, indicates not only that the provisions just described are highly effective, but that the general atmospheric conditions at night on the mountain will prove suffi- ciently good for this and even larger apertures. In this brief description of the new reflector and its accessories, I have called attention chiefly to those refinements and special features which, in designing the instrument, have appeared to me necessary. Mere bigness is no criterion of efficiency; if a great telescope is to yield a gain in results even approximately proportional to its increase in size, the utmost care must be given to meeting all those conditions which experience in the use of large telescopes has shown to militate against their successful performance. It was a most serious question whether it would be possible to give as fine a 81 12 G,W. RITCHEY figure to the 6o-inch mirror as was attained in the case of the 24-inch Yerkes mirror; the difficulties were of course incomparably greater, but the final figure of the 60-inch is decidedly better than that of the smaller mirror, and it is confidently expected that the temperature control will enable it to remain so while the telescope is in use. Simi- larly, it was a serious question whether the moving parts of the 60- inch telescope, weighing 23 tons (20,838 kilos), could be made to follow the stars as smoothly as those of the 24-inch, which weigh one ton (907 kilos) ; such smoothness of following is of course necessary for the finest results in photography; and it must be remembered that much greater smoothness and accuracy of motion are actually re- quired in the large instrument, on accoimt of its greater focal length and magnifying power. It is therefore a great satisfaction to see the star in the guiding eyepiece of the 60-inch remain perfectly bisected on the spider-lines for several minutes at a time, without perceptible tremor, and in addition, frequently to see the image of the guiding star itself, even with winter conditions, as small and sharp, and with its diffraction pattern as clearly cut, as I have ever seen in the 24-inch at Yerkes. For the successful performance of the new reflector, special credit is due to Mr. Jacomini, foreman of the instrument shop of the Obser- vatory; to Mr. Barnes, Mr. Schrock, and Mr. Kinney, for their great care and skill in the optical work, and to Mr. Dowd, for the installa- tion of the complicated electrical work of the telescope mounting and dome. I am indebted also to Director Hale for the opportunity afforded me to carry out in their entirety my plans for the great reflector and its dome, building, and accessories. January 7, 1909 8a I 1 i ft. \ .1 1 ' J ./I • * 1 •j I :1 •. N *!*• »» 1 # "•^ ~N (Zlanttgir fl^Mllil 01 ffp U ffm fl tftf^ icontributions from the mount wilson solar Observatory . NO; 87 ' •". < NOTE ON THE POLARIZING EFFECT OF COELOSTAT MIRRORS BY CHARLES E. ST. JOHN Reprinted from the Astrophysical Journal^ Vol. XXIX, Maiy 1909 .''^ * , *iJ » J * / J' / . I • I M k !' 1 Contributions from the Mount Wilson Solar Observatory, No. 37 Reprinted from the Astrophysical Journal, Wd. XXIX, pp. 30x-304, XQ09 NOTE ON THE POLARIZING EFFECT OF COELOSTAT MIRRORS By CHARLES E. ST. JOHN The discovery of the Zeeman eflfect in sun-spots by Hale' made it important to determine the action of the silver-on-glass mirrors of the tov^er telescope of the Mount Wilson Solar Observatory upon cirodarly polarized light. It was at once recognized that circularly polarized light would be changed to elliptically polarized by reflection from the silver surfaces, and that its eflfect would vary with the angles of incidence and hence with the position of the sun. A description of the tower telescope has been given by Hale' and it only needs be said here that the second mirror sends the light vertically downward through the 12-inch objective, and that about 4 feet below the objec- tive — focal length 60 feet — the beam was received upon the analyzer which served to fix the position of the axes of the elliptically polarized light and to determine their ratio. Light circularly polarized in a known direction was obtained by passing sunlight through a Nicol prism and then through a Fresnel rhomb whose plane of incidence formed an angle of 45° with the short diagonal of the Nicol. The direction in which it was necessary to rotate the rhomb to bring its plane of incidence into coincidence with the plane of vibration of the Nicol was taken as the direction of revo- lution of the circularly polarized light^ and it was distinguished as clockwise and counter-clockwise as viewed by the observer when looking in the direction of the source. This polarizer was mounted in a screen and interposed between the sun and the first mirror so that no light reached the mirror except that passing through the polar- izer. The analyzer consisted of a Fresnel rhomb mounted upon a divided circle, and a Nicol, likewise with a graduated mounting, that ' Contributions from the Mount Wilson Solar Observatory ^ No. 30; Astrophysical Journal, 28, 315, 1908. a Contributions from the Mount Wilson Solar Observatoryy No. 23; Astrophysical Journal, 27, 204, 1908. 3 Wood, Physical Optics, p. 277. 83] I 2 CHARLES E, ST. JOHN could follow the eccentric movements of the rhomb as the latter was rotated. Monochromatic light was obtained by interposing ruby glass in the train and its intensity was further reduced by means of a smoked glass. The observer lay on his back while making the observations which were carried out in the following way: The incident light was polarized in a clockwise direction and after reflection from the two mirrors fell upon the rhomb; a position of the rhomb was then found for which the Nicol could produce extinction of the light; the rhomb was rotated 90° and the position of the Nicol for extinction again determined; the operation was repeated with reversed polarization of the incident light. The readings of the scales give the azimuth of the plane of incidence of the rhomb and the corresponding azimuth of the long diagonal of the Nicol for extinction. The difference between the azimuths of the Nicol cor- responding to the first and second settings of the rhomb gives twice the angle between the major axis of the elliptically polarized light and the plane polarized light produced by the rhomb when its plane of incidence coincides with the major axis of the ellipse. This angle is given in the colunm of the table marked I. The ratio of the axes is of course given by the tangent of this angle. The difference between positions of the Nicol for clockwise and counter-clockwise light for the same setting of the rhomb is given in the column marked II, and approximates 90^. The great change in the position of the axis of the ellipse between morning and afternoon, as shown by the position of the rhomb, followed the transfer of the coelostat mirror from the west to the east side of the second mirror at noon. The condition of the silver sur- faces affects the results appreciably, but under normal working conditions the changes due to this cause are not great enough to be troublesome. In the practical application of these results in studying the Zeeman effect along the lines of force in the magnetic field of sun-spots, the rhomb was set with its plane lying northeast and southwest in azi- muth 20^ for the forenoon, in azimuth 70^ from noon to about 3 P. u., and then in azimuth 55^. The Nicol was set with its long diagonal in an azimuth equal to that of the rhomb plus 25^ in the forenoon and plus 30^ in the afternoon to extinguish clockwise light, and rotated 84 POLARIZING EFFECT OF MIRRORS s \ £ s £■ 3. - 'i s s, a 1 1 i \ I zzzzzzz •S-S-S'B'S'S'S zz "B'S 5ZZ ZZZZZZZ ZZZZ2ZZZZZZZZZZZ b O 00 O O P-03 m n O VI O "nco HgOi-rOQEiO O win OcQ i~vi ZZZZZZZZ2ZZZZZZZZZZZZZZZZZZZ ■8 "8 -S "S -S -8 -S -3 "S -5 -5 -S "S -5 -3 -S -5 -S -5 -S -S -S -3 -S -5 -S -3 -S II i :|| : :|| : iff i ifl : :|| i| ^S : :8g : :^g : :g| : :|| : :|| :| S'Sl'E jS't'f SS'S'taa'5'SsS'S'SSS't'iSS'fS UOOOOOOuoouoOouuuouuOouuOOOO 4 CHARLES E, ST. JOHN 90^ to extinguish Counter-clockwise light. Calculation of the effect produced by the plane of the rhomb not coinciding with the major axis of the elliptically polarized light shows that for a deviation of iQp the resulting effect is very elliptical light — ^axes in the ratio twelve to one — ^with the major axis forming very nearly the same angle with the plane of the rhomb as the plane polarized light which results when the plane of the rhomb coincides with the major axis of the elliptical light coming from the mirror. This allows some latitude in the setting of the rhomb provided the angle between the planes of rhomb and Nicol is maintained. Mount Wn.soN Solak Observatory February 1909 86 i I ' f A .•J c^ 5^Z0.3'SO iEwnxtQ\jt itiatitiitfim at VaaliUigtim - ■ ■ I II I I Contributions from the Mount Wii^on Solar Observatory ' NO. 8a A FURTHER STUDY OF THE H AND K LINES OF CALCIUM BT ARTHUE S. KING Reprinted from the Asirophysieai Journal, Vol. XXIX, June 1909 • .Ir I ,t J ^1:. -- 4 I '■~ ( --. ■^ •) '- ., 4 *- ^ !! I r ;> - s> -^^ *^- Ik ^ Contributions from the Mount Wilson Solar Observatory, No. 38 Reprinted fzoax the Astropkysical Jtmnud^ Vol. XXIX, pp. 581-389, 1909 A FURTHER STUDY OF THE H AND K LINES OF CALCIUM By ARTHUR S. KING The electric fiimace* in the Pasadena laboratory has been used for additional observations of the calcium spectrum by way of supple- menting the writer's former papers on this subject' The new fea- tures in the present investigation have been the use of higher dispersion in the blue and violet, giving more definite data concerning the width of the lines, and a comparison of the changes in the spectrum when the furnace is operated in vacutun, in hydrogen atmosphere, and in air at atmospheric pressure. The vertical Littrow spectrograph was used with the objective of 13 feet (4 m) focal length. The third order then gave a dispersion of 1.37 Angstrom xmits to the millimeter in the blue, permitting a good comparison of the widening of the spectrum lines. When an atmosphere of hydrogen was used in the furnace chamber, the gas was prepared from zinc and sulphuric acid and carefully purified. The furnace was pumped out to about 5 mm, flushed once or twice with hydrogen, and then filled with the latter to atmospheric pressure. The hydrogen was kept slightly above atmospheric pres- sure during the operation of the furnace. It was found necessary to use a different jacketing material about the furnace tube when hydrogen was employed, as the carborundum previously used, which gives very efficient heat insulation, appears to trap a certain amoimt of air when the chamber is pumped out. This air is given oflF when the car- borundum becomes heated and is at once removed if the pmnp is in connection, as is the case for work in vacuum. When hydrogen was used, however, the residue of air caused small explosions, besides contaminating the gas. The change made was to use two-inch graph- > Contributions from the Mount Wilson Solar Observatory, No. 28; Astropkysical Journal, 28, 300, 1908. * Astropkysical Journal, 27, 353, 1908; Contributions from tke Mount Wilson Solar Observatory, No. 32; Astropkysical Journal, 2S, 389, 1908. »7l I 2 ARTHUR S. KING ite blocks as jacketing material, these being cut to size and placed vertically side by side perpendicular to the furnace tube, a hole in each block allowing space for the tube. This jacketing was success- ful except that as heat insulation it proved far inferior to carborundum, the heat being carried away so rapidly that the current which usually gave nearly 3000° C. gave barely 2500°. The temperatures were suflBcient, however, to compare the effect of hydrogen with the other furnace conditions, runs being made for comparison with the graphite jacketing both with the furnace in vacuum and with air present, though the carbonmdum insulation was used with the last two arrangements to obtain the higher temperatures. The present investigation has verified in all essential points the conclusions arrived at in the previous papers, while the results with the furnace in hydrogen and in air throw interesting lights on the effects observed in the vacuimi experiments. The use of hydrogen at atmospheric pressure not only excludes oxidation, which is accom- plished well enough by pumping out the chamber, but alters the rate of diffusion of the metallic vapor in the tube and so affects the vapor-density. The loss of heat by convection was found to be very considerable with the hydrogen atmosphere and accounts in large measure for differences observed between these spectra and those obtained in vacuum with the same current through the tube. The presence of hydrogen in the furnace is very favorable to the fluted spectrum in the red referred to in papers by Olmsted* and the writer' and now being measured by the former from large-scale photo- graphs. This was observed during the present work both visually and by photographs made with red-sensitive plates. If these flutings arise from an actual compound of calcium with hydrogen, it would mean that hydrogen introduces a chemical action of its own. The formation of the compound, however, is by no means certain. To use the furnace in air, one of the outlet pipes was usually left open. As was to be expected, a lively combustion took place in the tube when air was present, although the tube itself was consumed but I Contributions from the Mount Wilson Solar Observatory, No. 21; Astropkysical Journal, 37, 66, 1908. 3 Contributions from the Mount Wilson Observatory, No. 35; Astropkysical Journal, 39, 190, 1909. 8 H AND K UNES OF CALCIUM 3 slowly and would last for several runs, there being no active circula- tion of air. There should have been more convection of heat than when hydrogen was used, but the oxidation more than balanced this loss, and brilliant spectra were obtained. At the higher temperatures the continuous ground soon became strong enough to give all lines in absorption. The assumption in the former papers, that the continu- ous spectrmn from the furnace is produced through the reflection of light from the white-hot walls of the tube by the suspended particles of vapor, appears to be correct; as when air was present and a high current was thrown on the tube, a series of exposures, each lasting a few seconds, showed that the mass of vapor which quickly arose gave a strong continuous spectrum in the earliest stages, while the bands were developing and before any lines appeared. With the furnace tube in vacuum or in hydrogen a large quantity of calcium in the tube as well as a high temperature was required to give the lines in absorp- tion. With the tube in air, the vigorous oxidation gave this condition with a very moderate supply of calcium. At least two runs of the furnace were made xmder each set of conditions, to eliminate effects of any irregularities in the action of the apparatus. Toward the end of the experiments a high-reading ammeter was installed, so that for some of the photographs the corresponding currents through the tube were measured. Pyrometer readings were frequently taken by sighting the instrument on as much of the wall of the tube as the small aperture of the furnace window permitted to be seen. These readings should give values close to the correct temperatures. Plate VIII gives a series of third-order spectra for different temperatures and imder the three conditions used during this work. A large portion of the spectnun between H and X 4227 is omitted from the plate. A ninnber of other photographs were taken, not always favorable for reproduction, which gave good comparisons of special conditions and will be referred to in the discussion. Furnace in vacuum. — ^The experiments with the furnace in vacuum were carried far enough to show that the higher dispersion fully confirms the conclusion of the former paper that H and K do not appear much below 2500° C. (perhaps 2400° is a better limit), that they are strengthened by higher temperatures but widen very slowly 89 4 ARTHUR S. KING within the range of temperature given by the furnace; while X 4227 appears at low temperature and depends for its width almost entirely on the amoxmt of vapor present. Plate VIII gives first an arc spectrum for the identification of the furnace lines. Furnace spectrum No. i is at about the limit of visibility of H and K as regards temperature but is made fairly strong through long exposure (30 minutes). The amount of calcium in the tube was small for both this and No. 2. A higher temperature gave No. 2, which was obtained with two minutes* exposure. The continuous ground is rather strong, but both tem- peratiure and vapor-density were low enough to prevent the spectrum passing into absorption. The relative strength of H and K in No. 12 compared to the group of arc lines near X 4300 is to be noted in con- nection with results under other furnace conditions. Furnace in air, — ^The use of air in the furnace chamber gave a very vigorous combustion and, if much calcium was present, a strong continuous spectrum. The image of the interior of the tube appeared brilliant white on the slit when a high temperature was employed. All lines and the cyanogen bands then appeared in absorption. Nos. 3 , 4, 5, 6, and 7 of Plate VIII show diflFerent conditions of the spectrum with air present. No. 6 is typical of the high temperature condition (2800® to 2900° C.) with strong vaporization. This condition was photographed several times during diflFerent runs of the furnace. The width of A 4227 could be made greater with more calcium present and of course appeared greater if the exposxure were so timed as to give a weaker continuous ground. This state is shown in No. 7, taken during another run, the continuous ground being weaker and the reversed lines wider. This was taken at an earlier stage of the run, when the amount of vapor was probably greater than for No. 6, the temperature being about the same (accurate pyrometer measure- ments were difficult during these short exposures). No. 5 was taken at lower temperature than either Nos. 6 or 7. Nos. 3 and 4 represent lower temperature conditions with air present, the spectrmn then showing bright lines. No. 3 was taken after No. 6 with an exposure twenty times as long as for the latter. The temperature for No. 4 was intermediate between Nos. 3 and 6, measuring about 2600° C. The amount of calcium was small for No. 4, the furnace having been used for a run the day before, and was 90 H AND K UNES OF CALCIUM S now heated without renewing the supply of calcium. The strength of the cyanogen bands in No. 4 is doubtless due to the tube being worn thin in parts, as it was now used for the sixth successive run. It is evident that H and K can be decidedly strengthened both in emission and absorption by increasing the temperature, especially when assisted by the strong chemical action given when air is present. As to how much they depend upon temperature can best be seen by comparing them with the group of arc lines from A 4283 to A 4319. These lines remain fairly narrow and change but slowly with increase of temperatxnre. Taking emission spectra, a comparison of Nos. 3 and 4 shows the K line somewhat weaker than X 4283 in the low-tempera- ture spectrxmi, while at high temperature K is decidedly the stronger. The same kind of plate (Seed's " 27 ") was used for all of these photo- graphs. Exactly the same relation is found for the lines in absorption by comparing the low-temperatiure absorption spectrum No. 5 with either No. 6 or No. 7. It is plain that H and K strengthen much more rapidly with increasing temperature than the average arc line. A comparison of Nos. i and 2 shows the same relation for the furnace in vacuum. Furnace in hydrogen. — ^The use of hydrogen in the furnace gave results which at first seemed to indicate that the presence of this gas suppressed the H and K lines. They were scarcely visible with the same currents through the tube which gave them very strong in vacuum and in air. Further investigation showed, however, that this eflFect was in all probability due solely to a lower temperature produced by the convection of the hydrogen at atmospheric pressure, combined with the absence of oxidation. Three runs of the furnace were made with vacuum, hydrogen, and air successively in the chamber, the graphite jacketing being used and all conditions of operation made the same as nearly as possible. Pyrometer measurements showed that for corresponding stages in the different rxms the temperatures were nearly the same for vacuum and for air, while for hydrogen they registered 150° to 200° lower. Another test was made by taking a photograph with hydrogen at atmospheric pressure, the temperatxnre being high enough to give H and K distinct though weak. This pho- tograph is reproduced in No. 8 of Plate VIII. The furnace was then pumped out and fresh hydrogen admitted to about 3 cm pressure, 91 6 ARTHUR S, KING after which No. 9 was taken with exposxire time and other conditions as nearly the same as possible. H and K show nearly as strong as would be expected in vacuum xmder the same conditions. The rapid change of H and K with temperature as compared to the arc lines near A 4300 is very clearly brought out in these two photographs, H and K being very much weaker than the arc lines in No. 8, while in No. 9 they are stronger than the strongest of the X 4300 group. The effect on A 4227 of change of vapor-density is very decided in Nos. 8 and 9. The temperature for No. 9 was considerably higher, but the furnace had partially cooled diuring the changing of the gas and the supply of calcium was doubtless much less than for No. 8. Two other photographs, not reproduced, give the same effect as Nos. 8 and 9 for a smaller temperature interval. Dependence of k 4227 on vapor-density. — ^All of the photographs support the view that the intensity of A 4227 is governed by the supply of calcium vapor. Visual observations also were very convincing. The spectnun was usually closely watched during the exposures by means of a direct-vision spectroscope of considerable dispersion. At the beginning of a rim, with fresh calcium in the tube, X 4227 was broadly reversed at moderate temperatures. It became narrower as time went on, and after a while the use of a much higher tempera- ture would not widen the line appreciably. The widening of a line of this character refers to the total width, not to the width of the reversal, the latter depending upon the distribution of heated vapor between the middle and ends of the tube. H and K in the arc and spark. — ^It may supplement the furnace results to consider the effects upon the H and K lines of conditions which are perhaps the most pronounced in their ability to alter the spectra of the arc and spark respectively. These are the change from the core to the outer layers or "flame" of the arc, and the use of self-induction with the condensed spark. The changes brought about by these influences are well illustrated by Plate IX. Nos. i and 2 are enlargements of prismatic spectra of the core and flame of the arc which were photographed by Dr. Olmsted in the Mount Wilson laboratory. The spectrograph was a Fuess instrument which gave very bright spectra, permitting short exposures, one-half second being used for the core and 15 seconds for the flame. A num- 9a H AND K UNES OF CALCIUM 7 ber of typical axe lines, such as the 4300 group and the series triplets at X 3624-44 and X 4426-55, are of nearly the same intensity in the two spectra. H and K are very much stronger in the spectnmi from the core of the arc. Nos. 3 and 4 of Plate IX are enlarged from a portion of the spark spectrum of calcium photographed by the writer with a one-meter concave grating in the University of Bonn and discussed in a paper' published at that time. No. 3 was taken with the strongly condensed spark and No. 4 with self-induction in circuit. The relative strength- ening of H and K as compared to the arc lines by the stronger discharge conditions is very decided. The enhancement of the pair of "spark lines" at XX 3706, 3737 is still greater than for H and K. Discussion. — ^A line of reasoning which can perhaps be developed into a working hypothesis seems to follow from the fact distinctly brought out by the furnace experiments, that the H and K lines show a very definite behavior when compared with representative arc lines, such as the group near X 4300 and the series triplet at X 4426-55 (shown on the plate with the previous paper'). These arc lines are of fair strength at a temperatiure just high enough for H and K to become visible. As the temperature rises, H and K increase in inten- sity (though without decided widening) much faster than do the arc lines, surpassing at the higher furnace temperatures the strongest lines of the X 4300 group (see No. 4 of Plate VIII) . If this rate of increase is maintained at higher temperatiu-es, the strength of H and K in the arc may easily follow. Although, as was remarked in the former paper, the difference between their intensities in the arc and furnace seems very large to be caused by the foxu: or five hundred degrees' difference in the measured temperatures of the furnace and the posi- tive terminal of the arc, we do not know even approximately the temperature which would correspond to the strength of the energizing influence in the arc which gives the light-vibration. To account on this hypothesis for the strong broadening of'H and K when a large amoimt of calcium is used in the arc, we should have to assume that this increases the combined electrical and chemical action of the I Astrophysical Journal, 19, 335, 1904. > CoiUribuHons from the Mount Wilson Solar Observatory, No. 33; Astrophysical Journal, a8, 389, 1908. 93 8 ARTHUR S. KING arc in a way that corresponds to higher temperature. The rapid increase of H and K when the furnace is operated in air shows to some extent how this may come about. It is then unnecessary to assume a dependence of H and K upon vapor-density, which in the furnace does not seem to hold. It is to be noted that the widening in the arc is a rapid shading-off from a strong central maximum, the intensity- curves for these lines being very different from that for X 4227. In both arc and spark the spectra in Plate IX show that the relative intensity of H and K increases rapidly with the strength of the dis- charge conditions and thus with the "electrical temperature." H and K have long been regarded as having to some extent the character of "spark lines," but the present work helps to bring out their relation to such typical "spark lines" as A 4481 of magnesium, ^ 5339> 5379 ^^ cadmium, and XX 4912, 4925 of zinc. The H and K lines show the same ready response as these lines to changing condi- tions in the source, but the range of temperature for their production extends much lower. It seems reasonable to suppose that temperature, if forced suffi- ciently high, could supply the energy to produce those conditions of the spectrum which can be given, in terrestrial sources, only by the electric discharge. It may then be possible to select a list of lines whose strength, relative to other lines, will serve in some measure to indicate the temperature corresponding to the radiating energy of any light-source, which often, as in the flame and electrical sources, would not agree with any thermal measurements to be made with our present instruments. H and K would have their place in such a list as being lines which first appear when the average arc lines have reached a considerable intensity and are changing slowly with increased tem- perature. Having once appeared, H and K are apparently very sensitive through a long range of temperature and are reinforced for this purpose later by the group of typical " spark lines," some of which require less energy of discharge for their production than others. A comparison with solar phenomena is of interest in this connec- tion. If the sensitiveness of H and K to increased temperature con- tinues at the solar temperatures, this would seem to be responsible in large measure for the great width of these lines in the spectrum of the body of the sun, their weaker condition in sim-spots, and thei 94 H AND K UNES OF CALCIUM 9 appearance at high levels in prominences, where the rarity of the vapor should not greatly affect their intensity if the temperature is high. Passing to stellar spectra, it may be said in general that H and K are strong in the spectra of those stars which show strong ''spark lines." These are the hotter stars if the "spark lines" furnish a valid basis of temperature classification. If their increase in intensity in some cases does not seem to be the same as for other '' enhanced lines," the possibility remains that H and K may lose their sensitiveness to some extent at extremely high temperatures, in the same way that we have seen arc lines lose their ready response much lower down. Mount Wqson Solar Observatory April 1909 95 , > •• ■, f ' 1 ♦ ■^i It ^Ci. 5^0. J ro .J Cfonif 9fe 9tuitititti0tt 0f VaBtffatgtott Contributions prom the Mount Wilson Solar Observatory NO. 8» THE ZEEMAN EFFECT FOR TITANIUM BY ARTHUR S. KING f Reprinted from the AUrophytieal J&umal, Vol. XXX, July 1909 :-:- .»_ » \ '. '••■^. >N^A^' , h U 14 M.'S. l-v^/^-<^r~ 1 \ n Contributions from the Mount Wilson Solar Observatory, No. 39 Reprinted from the Astrophysical Journal^ Vol. XXX, pp. 1-13, igog THE ZEEMAN EFFECT FOR TITANIUM By ARTHUR S. KING In a former paper^ the writer published a list of titanium lines with their separations in the magnetic field. These were lines which, for the most part, show a separation other than that of the normal triplet when the source of light is observed at right angles to the lines of magnetic force. They were selected from the mass of titanium lines to see how closely their types of separation agree with the sepa- rations of the same lines in sun-spot spectra. The comparison showed clearly that in the less refrangible part of the spectrum the agreement is so good as to leave little doubt that the magnetic field of the sun- spot is of such character that much of the light comes to us at a large angle to the lines of force. The photographs from which this study was made contain a more complete record of the behavior of the titanium lines in the magnetic field than has yet been published for the region from X 3900 to X 6600, so that it may be well now to publish a complete list of the measure- ments obtained for lines in this region. This list overlaps to some extent that published by Purvis, * but the majority of his measure- ments are for lines farther to the violet, only lines of very distinct separation being measured through the range covered by my plates. A large proportion of the titanium lines are either narrow triplets or of higher order of separation, so that even with the high dispersion used (about 0.93 Angstroms to the mm in the violet) quadruplets often appear as diffuse doublets, and other types have their components blended to such an extent that measurement is impossible. For this reason the Nicol prism was used very freely, when the light was observed across the lines of force, to separate the components given by light- vibrations parallel to the lines of force, the " transverse effect," I CofUrUnUions from the Mount Wilson Solar OhservaUfry, No. 34; Astrophysical Journal, 29, 76, 1909. a J. E. Purvis, Proceedings^ Cambridge Philosophical Society , 14 (i), 41. 97] I I 2 ARTHUR S. KING from those whose light-vibrations are in a plane perpendicular to the lines of force, the "longitudinal effect." The experimental method was briefly described in the former paper. The source of light was the electric spark given by a 5 K. W. transformer. The spark terminals were pieces of titanium held in brass clamps between the poles of a DuBois electro-mag- net. Self-induction was used as a rule in the spark circuit to sharpen the lines. The pole-pieces of the magnet were conical, of 60° angle, the faces being \ inch (6 mm) in diameter. The magnet was provided also with a pierced pole-piece which permitted the light to be taken parallel to the lines of force, along the axis of the magnet. Measurements of the field-strength were made by means of a bismuth spiral small enough to secure a good uniformity of field within its area. A considerable percentage variation of the field was found between the middle of the gap and the region close to the pole- pieces. This was shown also in the spectrum photographs by the bending outward of the ends of components of separated lines when the slit was parallel to the lines of force and rather long. A magnetic gap of -jfij incli (8 mm) was generally used and there was little variation of the field over the middle half of the gap where the brightest portion of the spark played which was focused on the slit. The majority of the photographs were taken with a field-strength of 12,500 gausses. Other field-strengths of 13,800 and 18,400 were o used. With the latter field a second-order dispersion of i . 37 Angstrom units to the millimeter was usually ample to determine the character of the separated lines. The photographs were all made with the 13-foot (4 m) vertical Littrow spectrograph. The second and third orders were used for all photographs on which final measurements were based; though the first order was occasionally employed to get a general view of a large region. A small metal platform above the slit supported a Nicol prism in a rotating holder. By rotating the latter through 90°, either, of the polarized components of the light from the spark could be allowed to pass. In the following table, the measurements are for a field of 12,500 gausses, the separations for stronger fields being reduced to this value. 98 THE ZEEMAN EFFECT FOR TITANIUM 3 The order of separation which the line undergoes in the magnetic field is given in the second column, with the reservation that further separation of the components might appear with still higher dispersion. This would occur for the most part in the case of the more complex types. Thus it is possible that some of the lines given as quadruplets would show their outer components still farther separated. The same is true for the quintuplets, some of which probably have at least seven components. The third column, headed "longitudinal effect," gives the separations observed when the light is viewed along the lines of mag- netic force. Sharper lines were obtained by viewing the spark across the lines of force and taking the light from the middle of the gap. Most of the measurements for the longitudinal effect were obtained in this way, the Nicol prism being turned so as to transmit the light with vibra- tions in a plane at right angles to the lines of force. The photographs for the fourth column, giving the " transverse effect," were then obtained by turning the Nicol through 90°, thus transmitting the light with vibrations parallel to the force-lines. The numbers give the separa- tions of any components that could be measured under these condi- tions in terms of AA/A*, the number given to be multiplied by 10"*. There are a few lines for which the order of accuracy in the measurements is considerably less than for the main body of the lines, due to blends or other disturbing features in the photographs. These lines are denoted by an asterisk. Disctissian of tables. — An examination of the tables shows that while the great majority of the titanium lines are separated into triplets by the magnetic field, other types of separation are well repre- sented. The largest number of components observed were eight for the line A 4281 . 53, this line showing five components for the longi- tudinal effect and three for the transverse, the constituent lines being of about equal intensity in both cases. Separations of seven, six, five, and four components are each represented by a group of lines. The relative intensities of the several constituents for lines of these types are often very different, as is the case in other spectra. From the point of view of regularity of structure, the most interr esting lines are perhaps the two of seven components each, A 4527.49 and A 4544.86. These lines are duplicates as regards separations. Arranged in the conventional way, the minus sign denoting the violet 99 ARTHUR S. KING Spark 1 Observed at AX/X- Spark. AX/X- Spark, OAtmavka A Right Anfflfs to Force-Luies Longitudiiul Effect Transverse Effect 3904.93 Triple At least 1. 19 Single 3921 • 56 Components 1-73 quin- blended, tuple probably triple 3924.67 Quadruple 1.47 0-78 3926.46 Triple At least Z.28 Single 3930.02 Components 1.36 quin- blended, tuple probably triple 3932.16 Triple 1.94 Single Enhanced line 3934.37 Triple 1.47 Single 3947 92 Triple 0.78 Single 3948.82 Triple 1.04 Single 3956.48 Triple Z.ZI Single 3958 36 Triple 1.56 Single 3962.99 Triple 2.04 Single 3964.42 Triple 1.88 Single 3981.92 Triple 1.67 Single 3982.63 Sextuple Inner 0.80 Outer 2.87 2.22 A ■ • • 3989.91 Triple 1. 41 Single 3998.79 Triple 1.67 Single 4002.65 Triple 1.35 Single 4003.91 Triple 2.04 Single 4009.08 Triple 1.65 Single 4009.81 Unsepa- ■ rated * » • • * • • ■ 4012.54 Quadruple 0.99 0.81 Longitudinal compo- nents widened. Perhaps narrow sextuplet. En- hanced line 4013.80 Triple Z.22 Single 4021.89 Triple 1. 17 Single 4024.73 Triple 1.82 Single *4025.29 Triple ? 1-39 Widened, not resolved Enhanced line 4026 . 69 Triple 1.26 Single 4028 . 50 Triple 1.29 Single Enhanced line 4030.65 Triple ^'33 Single 4034.05 Triple 1-53 Single 4035 98 Triple 1.62 Single 4053.98 Triple 1.28 Single Enhanced line 4055-19 Triple 1.79 Single Enhanced line 4060.42 Triple 1.84 Single 4064.36 Triple 1.70 Single 4065.24 Triple 1.84 Single 4078.63 Triple 1. 81 Single 4082.59 Triple 1. 81 Single 4109.95 Triple 1-47 Single 4113.87 Quadruple 1.48 X.21 4x22.31 Triple 1.25 Single 100 THE ZEEMAN EFFECT FOR TITANIUM ^ Spark A Observed at Rurfit Angles to Force-Lines AA/X» Spwrk, Longitudinal Effect AA/A> Spark. Transverse Effect Remarks 4123.71 Triple 1.28 Single t 4127.69 Triple 1.38 Single 4129-34 ? • » • • • • • • Components faint, ap- parently triple in longi- tudinal efifect 4137.43 Triple Z.64 Single 4151-13 Triple 1-34 Single 4159-80 Triple 1. 16 Single 4161.68 Triple 2.05 Single Enhanced line 4163.82 Triple 1. 41 Single Enhanced line 4169.46 Triple 1-37 Single 4171.21 Triple 1. 18 Single 4172.07 Triple 1.22 Single Enhanced line *4i73-62 Triple 1.69 Single Enhanced line 4186.28 Triple 1.24 Single 4188.84 Triple 1-77 Single 4201.95 Triple 1.83 Single 4203.94 Triple I 77 Single 4238.00 Triple 1.24 Single 4256.29 Triple 1.99 Single 4260.99 Triple 2.20 Single 4261.75 Triple 1.79 Single 4263.29 Triple 1.42 Single 4272.70 Triple 2.31 Single 4274.75 Triple 1.20 Single ■ ■ • • 4281 . 53 Octuple Inner jo.oo (0.81 (1-83 Outer ^0.00 (i-77 ■ • ■ • • • • • 0.81 0.00 0.91 4282.86 Triple Z.02 Single 4285.16 Triple 2.29 Single 4286.17 Quadruple 1-73 0.70 4287 . 57 Quadruple 1.82 0.63 ♦4288.04 Triple ? 1. 81 Single ? M'dle compon't widened 4289.24 Quadruple 1-59 1.09 4290.38 Triple 1.36 Single Enhanced line To. 71 ■ • • • Blend of red component 4291. I I Quintuple jo.oo 1.77 with unseparated line (0.64 • • • • may cause apparent asymmetry of longi- tudinal triplet 4291 . 28 Unsepa- rated • ■ • • • * « • 4294 . 20 Triple I-5I Single Enhanced line 4295-91 Unsepa- rated • • • • • • • ■ 4298.83 Triple 0.97 Single 4299.41 Triple 1.03 Single ♦4299.80 Triple 1.64 Single 4300-21 Triple Blended with adjacent lines Single Enhanced line lOI ARTHUR S. KING Spark X Observed at Right Anfflcs to Force- Unea AA/A" Spwk, T/mgitudiiuil Effect AX/A« Spark. Transverse Effect Remarks 4300 -73 Triple 1. 10 Single 4301 . 16 Triple 1.37 Single 4302.08 Sextuple Inner 0.67 Outer 2.47 0.93 Enhanced Une 4306.08 Triple 1.52 Single 4308.64 Quadruple 1.89 0.99 4313 03 Quad- ruple? 1.85 Widened, not resolved Enhanced line - 4314.96 Triple 1.65 Single Enhanced Une j ^^'.^nd 431.S14 Quadruple 1-55 I.I3 4318.82 Triple 1.39 Single ♦4321.12 Sextuple Inner 1.17 Outer 3. II 1.05 *432i.8i Triple 1.26 Single 432S-3I Triple 1.28 Single 4326.52 Triple I 63 Single 4338 08 Triple 1. 18 Single Enhanced line 4344-45 Triple 1.88 Single Enhanced line 4351 00 Triple 1.48 Single Enhanced line 4367 84 Triple 1-30 Single Enhanced line 4387.01 Triple 1. 17 Single Enhanced line 4399-94 Triple 1. 17 Single Enhanced Une 4404 -43 Triple 1.39 Single 4417-45 Triple 1.36 Single 4417.88 Triple 0.90 Single Enhanced line 4418-50 Triple 1.46 Single 4421.93 Triple 1.31 Single Enhanced line 4426 . 20 Triple 1.20 Single 4427 - 27 Triple 1. 19 Single 4434-17 Triple 0.98 Single 4440.52 Quadruple 1.24 0.84 4443 98 Triple I.I5 Single Enhanced line ♦4444.73 Quadruple 1.02 0.96 4449-31 Triple 1-44 Single 4450-65 Sextuple 2.46 1. 00 Longitudinal effect is mean for two outer pairs. Enhanced line 4451.09 Triple 1.27 Single 4453-49 Triple 0.78 Single 4453.88 Triple 0.87 Single 4455-48 Triple 1.30 Single 4457.60 Triple 1-53 Single (1.08 • • * • Central line very strong 4464.62 Quintuple \ 0.00 I 05 in longitudinal effect ( 1.02 * • • • Enhanced line 4465.98 Triple 1.74 Single 4468.66 Triple 1-39 Single Enhanced line 4471.02 Quadruple Too weak to measure 1-54 4471-41 Septuple ? Inner pair 1.43; outer pair too weak to measure Components blurred, probably triple 102 THE ZEEMAN EFFECT FOR TITANIUM Spark X. Observed at AA/A- Spark, AA/X- Spark, Remarks #% Right Angles to Force-Luies LongitudiQal Effect Transverse Effect 4480.75 Triple 3-04 Single 4481.44 Triple 2.00 Single 4482.84 Triple 1-93 Single 4488.49 Triple X.29 Single Enhanced line 4489.26 Septuple ? 1.53; also, a taint outer Components blurred, pair probably triple 4496.32 Triple 1-77 Single 4501-45 Triple 1. 19 Single Enhanced line 45". 91 Triple i-75 Single 4518.20 Triple 1.83 Single 4522.97 Triple 1.88 Single 4527 -49 Septuple Inner z.15 Outer 2.46 ( 0.60 j 0.00 (0.60 4529.66 Quadruple Z.IO 0.84 Enhanced line 4533-42 Triple 1.76 Single 4534 . 14 Triple 1.38 Single Enhanced line ♦4535-74 Triple 1.72 Single ♦4536.09 Triple 1.70 Single 4536.22 Unsepa- rated • • • « • • ■ • Line has close companion to red which appears unseparated 4544.86 Septuple Outer 2 . 39 Inner too faint to measure ro.6o •) 0.00 (o.6i 4548.94 Triple 1.86 Single 4549.81 Triple 1-39 Single Enhanced line 4552.63 Triple Z.84 Single 4555.66 Triple 1. 81 Single 4563 -94 Triple 1.04 Single Enhanced line 4572.16 Triple 1. 21 Single Enhanced line 4590-13 Quadruple 1.22 0.97 Enhanced line 4617.45 Triple 1.47 Single 4623.28 Quadruple 1-37 Not clearly re- solved 4629 -53 Septuple ? Too weak to measure (0.55 i 0.00 (0.54 4638.05 Triple 1.86 Single 4639 • 54 Quadruple 2.12 0.76 ♦4639 85 Quadruple Components blended with adjacent lines 0.64 *4640.i2 Quadruple 2.75 1. 18 4645 -37 Triple 3.08 Single 4650.19 Triple 2.54 SiAgle 4656.64 Triple 1. 00 Single 4667.77 Triple 1. 19 Single 4682 . 99 Triple 1.27 Single 4691.52 Triple 1.42 Single 103 i 8 ARTHUR S. KING Spark X Observed at Right Angles to Force-Unes AA/A- Spark, Langitudinal Effect AA/X" Spark. TraDsvcrse Effect Remarks 4698.9s Quadruple 1-33 037 4710-37 Quad- ruple? Triple 1.76 Not clearly re- solved Components very hazy 4743.98 0.84 Single 4758-31 Triple 1.24 Single 4759-46 Triple 1-43 Single 4792.70 Triple 1. 14 Single 4799.98 Quad- ruple ? Triple 1.06 Not clearly re- solved 4805 ."28 1-33 Single Enhanced line 4805.61 Triple 1.30 Single 4820.59 Triple 1-23 Single 4841.07 Triple 1-25 Single 4856 . 20 Triple 1.35 Single 4868.45 Triple 0.95 Single 4870.32 Triple Z.18 Single 4885.26 Triple 1.36 Single 4900.10 Triple 1.27 Single Widened, not 49" -37 Triple? 1-39 resolved 4913.80 Triple 1. 12 Single 4920.05 Triple 113 Single 4921.96 Triple 1-44 Single 4928.51 Triple 0.91 Single 4981.91 Triple 1.49 Single 4991.25 Triple 1.38 Single 4999-69 Triple 1.28 Single 5001.16 Triple? 125 Hazy 5007.40 Triple 1. 01 Single 5013-48 Triple 1.46 Single 5014-37 • • ■ ■ • « • • • » ■ k Components blended with foreign line 5016.34 Quadruple 1.69 0.81 5020.21 Quadruple 1-52 0.86 5033-05 Quadruple Z.48 0.95 5025-03 Sextuple 0.43 and two hazy outer components 1.63 5025-75 Triple 1.62 Single ♦5036.09 Triple 1. 41 Single ♦5036.64 Triple Z.26 Single 5038.58 Triple Z.02 Single 5040.14 Triple 1.28 Single 5064.84 Triple 1.36 ' Single 5066.17 Quadruple 0.88 1. 10 5071.66 Triple 1. 41 Single 5113.62 Triple 1.27 Single 5120.59 Triple 1-33 Single 5129.34 Triple 1.24 Single Enhanced line 5145-64 Triple 1-43 Single 5M7-65 Triple 2.07 Single 5152.36 Triple Z.92 Single 5154-24 Triple X.92 Single Enhanced line 104 THE ZEEMAN EFFECT FOR TITANIUM Spu-k X Observed at AA/A" Spark. AA/X" Spark. Remarks r\ Risht Angles to Force-Unes Longitudinal Effect Transverse Effect 5173-92 Triple 0.88 Single 5186.07 Triple I.I3 Single Enhanced line 5188.86 Triple 1.51 Single Enhanced line 5193 14 Triple 1.38 Single 5206.22 Triple 1.45 Single 5210.56 Triple 1.59 Single 5212.50 Triple 1.51 Single 5219.88 Triple 2.21 Single 5222.85 Unsepa- rated • • • • • • • ■ 5223.79 Triple 1.56 Single 5224.47 Triple 1.80 Single 5224.71 Triple 1.52 Single 5225.20 Triple 1. 01 Single 5266.14 Triple 1.50 Single 5283.61 Triple 1.31 Single 5297.41 Triple 1.25 Single 5298.67 Triple 1.33 Single 5336.97 Triple 1.31 Single Enhanced line 53SI.26 Triple 1. 10 Single 5381.22 Triple Z.12 Single Enhanced line 5397.82 Triple 1. 16 Single 5409.82 Triple 1.28 Single 5418.98 Quadruple 1.34 0.99 Enhanced lin^ 5426.47 Quadruple 1. 81 1.38 5429.3s Triple 1.87 Single 5453.86 Triple 1.40 Single 5460.72 Quadruple 1.90 1.44 5471.41 Triple 1-43 Single 5472.92 Triple 1.42 Single 5474.44 Triple 1.32 Single 5477.90 Triple 1.48 Single *5482.o8 Quadruple 1-43 1.43 Given as 5281 .65 in for- mer paper. Measure- 5490.37 Triple 1.22 Single ment difficult 5504.12 Triple 1.27 Single 5512.01 Triple 1-55 Single 5512.74 Triple 1.46 Single 5514-56 Triple 1. 00 Single 5514.75 Triple 1.24 Single 5644.36 Triple 1.38 Single 5648.80 Triple 1.40 Single 5662.37 Triple 1. 41 Single *5663.i6 Triple 1.13 Single 5675 65 Triple 1.41 Single 5689.69 Triple 1. 21 Single 5702.88 Triple 0.90 Single 5708.46 Triple 1.72 Single 5712.09 Quadruple 1.68 0.9^ 5714.12 Unsepa- lated ■ ■ • • ■ • • • 5715.31 Triple 1-43 Single 5716.67 Quadruple 1.28 0.78 105 lO ARTHUR S. KING Spark k Observed At Right Ancles to Force-Uncs AX/A> Spark, Longitudinal Effect AA/A> Spark, Transvene Effect Remarks (103 • • • 5730.67 Quintuple j 0.00 1.93 *574o.2o Triple \ *'04 1.09 ■ • • • Single ♦5740.82 Triple 1.25 Single 5766.55 Triple 1.24 Single 5774.25 Triple 1-34 Single 5781 13 Triple 1.50 Single 5786.19 Triple Z.60 Single 5804.48 Triple 1.52 Single 5866.68 Triple 1.05 Single 5899.52 Triple 1.42 Single 5903.56 Triple 1-74 Single 5918.77 Triple 1-77 "Widened, not Single 59". 33 Triple Single resolved 5938.04 Quadruple 1-74 Widened, not resolved Z.29 5941.98 Quadruple 1.52 Longitudinal components very hazy 5953.39 Triple Z.26 Single 5966.06 Triple 1. 19 Single 5978.77 Triple 1.02 Single ♦5999.92 Triple 1.68 Single 6064 . 85 Triple 2.38 Single 6085.75 Quadruple 2.18 0.61 6091.40 Triple 1.29 Single 6126.44 Triple 1-37 Single 6215.46 Triple 1-34 Single 6258.32 Triple 1.68 Single ) Single ) Blend makes measure- 6258.93 Triple Z.IO ments difficult 6261.32 Triple Z.XI Single 6303.98 Quadruple 1.24 1.23 6312.46 Quadruple 1.54 0.92 6336 -33 Triple 115 Single 6366.56 Triple 1-15 Single 6546.48 Triple 0.97 Single 6554.47 Triple 1-32 Single 6556.31 Triple 1.54 Single component, and s and p the longitudinal and transverse components respectively, we have 4527.49 -1.23 s —0.60 p -0.58 s 0.00 p +0.57 s +0.60 p + 1.23 5 4544 . 86 —1,19 s —0.60 p • ■ • • O 0.00 p -h. . . . 5 -|-o.6i p + 1 .20 s Z06 THE ZEEMAN EFFECT FOR TITANIUM 1 1 The intervals of the side components from the central line are in the ratio 1:1:2. Such a close similarity would seem to be based on a series relation, but no other members of the series have been found. The line A 4281.53, of eight components, has its separations also approximately in the ratio 1:1:2, but the values of AA/A' are quite different from those of the above lines, and there is a central line for the longitudinal effect. The difference in wave number for ^4527.49 and 4544.86 is 84.4. That for the two sextets AA 4302.08 and 4321.12 is 88.9. The separations for these lines are as follows: —1.23 5 /— 156 ^ —0.46 p I —0.58 s -0.34 s \-o.53 p 4302.08 ( 0.00 4321.12 / 0.00 -I-0-33 ^ )-l-o-S3 P -1-0.47 P I +O.S9 s -hi. 24 s \ +1.55 s giving the ratio 2 : 3 : 8 for the first and i : i : 3 f or the second. Three lines of five components each which show as triplets when viewed along the lines of force are also of interest: 4291 .11 -0.89 p / — 1.08 s / —1.03 5 —0.71 5 (-053 p \ —0.96 p 0.00 S 4464.62 / 0.00 s 5720.67 < 0.00 5 -f-0.64 S /-f-0.52 p /-ho. 97 p -1-0.88 p \ -hi. 02 5 \ -hi. 04 s The ratios for these lines are approximately 2:3 (see note in table) 1:2, and 1:1 respectively, the arrangement of components for the first being different from that of the other two. The lines having four components show little regularity in their separations. The thirty lines which are most clearly of this type show five with the ratio of 5:4 for the longitudinal and transverse pairs, five with the ratio of 2:1, four with the ratio of 3:2, four with the separations approximately equal. A number of other ratios are represented each by one or two cases. Only one quadruplet, A 5066.17 shows a larger separation for the transverse pair than for the longitudinal. The quintuplets A 4291. II and of A 4464.62 give triplets in the 107 12 ARTHUR 5. KING longitudinal effect. A test was made of the polarization of these lines by taking the light through the axis of the magnet and passing it through a Fresnel rhomb and Nicol prism. Photographs were made for two positions of the Nicol 90° apart. The red and violet com- ponents of the longitudinal triplets were then successively extin- guished by the same positions of the Nicol which quenched the corresponding components of the other lines (appearing as doublets or quadruplets for this position) ; while the middle line of the triplet was unaffected by the change. Those titanium lines which occur in Lockyer's list of "enhanced lines" are noted in the "Remarks" colunm. These lines do not appear to be in a class by themselves as regards the types of separation which they undergo in the magnetic field. Of the 44 enhanced lines in the above table, 35 are triple, 6 are quadruple, i quintuple, and 2 sextuple. Many of the "enhanced lines" show but narrow separa- tions, especially those for which the difference between the intensity in arc and spark is most decided. However, the average separation of the 35 triplets is the same as that of the 195 triplets which are not " enhanced lines," the mean value of -r^ for each group being 1.46. Description of plates. — Plate X shows a remarkable group in the blue, near A 4300, which includes lines showing several types of separation. The plate was made by placing two photographs, made with the spark observed at right angles and with the Nicol prism in two positions 90® apart, side by side, each photograph showing a comparison spectrum of the spark without the magnetic field Beginning with the more complex, A 4281 . 53 has eight components. A triplet shows faintly in the reproduction for the transverse effect; while a stronger photograph gives five components for the longitudinal effect. We then have A 4302.08 of six components. This line shows a doublet in the transverse effect The opposite arrangement has not been observed in this spectrum. The quin- tuplet A 4291. 1 1 is of the type showing a triplet in the longitudi- nal effect. Its red componentf orms a blend with the single line A 4295. 28. Quadruple lines of varying degrees* of separation are seen in this group. The transverse effect gives the narrower pair, a condition 108 i ^ z I ^=1 S 5 S S. ■I THE ZEEMAN EFFECT FOR TITANIUM 13 which is general throughout the spectrum, though some lines in the yellow and red show nearly the same separation for both pairs. There are a number of the normal triplets, including the very wide A 4285. 16. A good example of unseparated line is A 4295.91, which shows equally narrow for both positions of the Nicol. The single line X 4291 . 28 which forms a blend has been noted. Plate XI shows, beginning at the left, the wide triplets A 4518.20 and A 4522 . 97, the arrangements of the photographs with comparison spectra being the same as for Plate X. Next comes the interesting line A 4527.49, of seven components, which has as a magnetic dupli- cate A 4544.86. The next group is a noteworthy collection in the yellow, the five lines to the right showing different kinds of separation. We have first (second line from the left) the narrow quadruplet A 5712.09, the separation of the transverse component being barely measurable, the line A 5714 . 12, apparently unaflFected by the magnetic field, the strong triplet A 5715 . 31, the fairly wide quadruplet A 5716 . 67, and the quintuplet A 5720.67, this latter showing as a triplet when observed along the lines of force, of the same type as A 4291. 11 in Plate X. A 5941 . 98, the last line at the right of Plate XI, is a type of quadruplet which has its longitudinal doublet of wide and hazy components which nearly fill up the central space. This type has interesting analogies in sun-spot spectra, as have several lines in the previous group (near A 5700) which were discussed in the previous paper. I am indebted to Miss Wickham for a large amount of careful work in the measurement of the photographs and the reduction of the values obtained. Mount Wilson Solar Observatory May 1909 109 \ ■' I . " \ k* '■ I« i This book should be returned to the Library on or before the last date stamped below. A fine is incurred by retaining it beyond the specified time. Please return promptly.